You are on page 1of 98

University of the Philippines Diliman

School of Economics

An Analysis of the Multidimensional Energy Poverty Index:

The Philippine Case

A thesis submitted

in partial fulfilment of the requirements for

Econ 199

Alyssa Mari G. Catanjal

Trexia Mae D. Lechoncito

27 May 2017

i
An Analysis of the Multidimensional Energy Poverty Index:
The Philippine Case

Alyssa Mari G. Catanjal and Trexia Mae D. Lechoncito

ABSTRACT

Despite the fact that there were studies that show that improving energy access contributes to
poverty alleviation, there have only been a few studies here in the Philippines that look at the status
of energy poverty. To this, we aim to introduce energy poverty in a multidimensional approach.
Through this approach, a more in-depth and targeted way can be made in order to assess the
situation and solve the problem at hand. We used the Household Energy Consumption Survey
(HECS) of 2004 and 2011 which has sufficient sample households to represent the Philippines.
Data from HECS include levels and patterns of energy consumption such as electricity access,
usage of modern energy sources, and ownership of recreational appliances. We adapted the Alkire-
Foster methodology in construction of the Multidimensional Energy Poverty Index (MEPI), which
captures both the incidence and intensity of energy poverty, to assess the extent of energy poverty
in the Philippines. Using this method, the results show that MEPI exhibits a significant reduction
from 0.611 to 0.552 in 2004 to 2011 as fueled by a large decrease both in the headcount and
intensity ratio. Moreover, the population of the MEPI-poor is heavily concentrated in the rural
areas and the Mindanao region particularly in ARMM, Zamboanga Peninsula, and
SOCCSKSARGEN where the use of traditional fuels for cooking is prevalent.

Keywords: poverty measure, poverty, energy sources, energy

JEL Classification: I32, Q40

ii
ACKNOWLEDGEMENTS

We would like to extend our heartfelt gratitude to our thesis adviser, Dr. Majah-Leah Ravago for

her infallible guidance and effort throughout the semester. We also want to thank Dr. Arsenio

Balisacan whose paper was used as the main model of this paper, and Dr. Gaurav Datt for

entrusting us a copy of his research that is pertinent to this study. This paper wouldn’t also be of

existence if it weren’t for the technical support Mr. Jan Carlo Punongbayan has provided us from

time to time, and for the feedback and assistance of Dr. Dennis Mapa, Dr. Karl Jandoc and Prof.

Manuel Leonard Albis. Furthermore, we want to thank Ms. Shirra de Guia of the Energy Policy

and Development Program (EPDP) and Mr. Alex Mañago of Philippine Statistical Authority

(PSA) for providing us with relevant data information, and the UPSE library and staff who guided

us in the pre-stages of this thesis.

To our family especially our parents, Felix and Amalia Lechoncito, and Maria Lourdes Catanjal,

we are very much grateful for your patience and understanding during the course of time when we

were doing our thesis. To our friends who showed undying support, we are thankful. Thank you

for instigating and encouraging us to finish the tasks at hand. To God be the glory!

iii
HONOR STATEMENT

We attest that this thesis we have submitted is our own. We have not cheated, plagiarized, nor

receive unauthorized assistance in the completion of this paper.

We have obtained the required prior consent for the use of the data for this research.

We understand that the University of the Philippines may impose commensurate sanctions and

penalties for instances of academic dishonesty committed in the completion of this paper.

Alyssa Mari G. Catanjal Trexia Mae D. Lechoncito

iv
WAIVER

Relevant portions of this work may be quoted and used for research and other scholarly purposes,

provided the proper citation is made.

Alyssa Mari G. Catanjal Trexia Mae D. Lechoncito

v
TABLE OF CONTENTS

CHAPTER 1: Introduction ………………………………………………..…….…..…….…… 1

CHAPTER 2: Trends of Energy Consumption …………………..………….……...………… 4

2.1. Trends of Energy Consumption in the Philippines ………………..……………... 4

2.2. Energy consumption in the Philippines vs. other Southeast Asian countries ….. 6

2.3. Perceptions on the Use of Common Household Energy Sources in the Philippines
…………………………………………………………………………............................. 8

CHAPTER 3: Review of Related Literature ………….………..…...….……..........................10

3.1. Definition of Energy Poverty …………………………………………………........10

3.2. Studies on Energy Access and Energy Poverty …………………………………...10

3.2.1. Studies that used Energy Access in the Philippines …………………… 11

3.2.2. Cross-country studies on Energy poverty ……………………………... 12

3.2.3. Empirical Studies using Alkire-Foster (AF) Model in the


Philippines …………………………………………………………..…………. 13

3.3. The Role of Energy Access ………………………………………………..…….… 14

CHAPTER 4: Existing Energy poverty metrics …………………………………...……........ 18

CHAPTER 5: Conceptual Framework ……………………………………………………..... 25

CHAPTER 6: Data and Empirical Approach ………………………………………...……... 29

6.1. Data ………………………………………………………………………………... 29

6.2. Summary Statistics ………………………………………….……………………. 30

6.3. Empirical Approach ………………………………………….…………………... 31

6.3.1. Multidimensional Energy Poverty Index (MEPI) ….…………………. 33

CHAPTER 7: Trends, Analysis, and Discussion of Results ………………………………... 39

7.1. Trends of MEPI Indicators …………………………….……………….…...…… 39

vi
7.1.1. By Urbanity …………………………………………….……………….. 39

7.1.2. By Region ………………………………………………..……………..... 40

7.1.3. By Industry Sector and Gender……………...……….…………..…….. 42

7.2. Results and Discussions ……………………………………….………………….. 42

CHAPTER 8: Conclusion and Policy Implications ……………………...……………………56

BIBLIOGRAPHY ……………………………………………………...…….……………….. 58

APPENDIX

A. Tabular Values of Data and Results …………………………….…...…………… 65

B. Stata Runs ………………………………….………………….…...……………… 75

B.1. MEPI Computation …………………………….…………………………75

B.2. Regression Results …………………………………..……………………. 78

B.3. Correlation Tests Results …………...………………..……………...….... 81

C. Documentation ………………………………………………..…………………… 84

vii
LIST OF TABLES

Table 2.1. Electricity access in Southeast Asia, 2013 ………………………..……......….…… 7


Table 2.2. Population relying on traditional biomass for cooking, 2013………….…....…..... 8
Table 3.1. Summary of Empirical Studies on Quantifying Energy Access in the Philippines
………………………………………………………………………………………………..…. 11
Table 3.2. Summary of Empirical Studies on Energy Poverty ……………..………………. 12
Table 3.3. Summary of Empirical Studies that used AF Model in the Philippines ……....... 14
Table 4.1. Dimensions, Indicators and Variables used for MEPI ……………………..……. 24
Table 6.1. Summary Statistics of Select Household Characteristics ……………..…………. 30
Table 6.2. Summary Statistics of Household’s Average Monthly Income (in cash and in kind;
2004 and 2011) …………………………………………………………………………………. 30
Table 6.3. Energy dimensions, 2004 and 2011 ……………………………………..…………. 31
Table 6.4. Description of dimensions, weights, and their cut-off points for computing MEPI
………………………………………………………………………………….…………….…. 34
Table 7.1. MEPI and its components H and A ……………………..……………………..….. 43
Table 7.2. Breakdown of MEPI into its dimensions ………………….…………………...…. 43
Table 7.3. Energy poverty profile by urbanity, 2004 and 2011 ………..………………….… 44
Table 7.4. Energy poverty profile by region, 2004 ………………………...…………………. 46
Table 7.5. Energy poverty profile by region, 2011 …………………………...………………. 47
Table 7.6. Tests for correlation results of MEPI ………………………………………........... 50
Table 7.7. Contribution of energy dimension to MPI, 2003 and 2012 ……………………… 53
Table 7.8. MEPI under different k values …………………………………………….…….... 55

viii
LIST OF FIGURES

Figure 1.1. Energy consumption per capita in Southeast Asia, 1990-2013 ………….………...1
Figure 1.2. ASEAN’s energy access and security in EAPI (2013-2017) ……………….……....2
Figure 2.1. Average energy consumption by sector in the Philippines (1998-2015) …….….. 4
Figure 2.2. Type of fuel used by households in the Philippines, 1989-2011 (in %) …………. 5
Figure 2.3. GDP per energy use (constant 2011 PPP $ per kg of oil equivalent) ..………….. 7
Figure 2.4. Perceptions of households on the use of common household energy sources, 2004
and 2011 ……………………………………………………………………………..................... 9
Figure 5.1. Energy poverty line ………………………………………………………………...26
Figure 5.2. Energy access multi-tier framework …………………………………………….. 27
Figure 5.3. Steps from energy availability to poverty reduction ………………………......... 28
Figure 7.1. Households Without Access to each MEPI indicators by urbanity, 2004 and 2011
……………………………………………………………..………………………...………….. 40
Figure 7.2. Households without access to each MEPI indicators by region, 2004 and 2011
…………………………………………………………………………………………..………. 41
Figure 7.3. Contribution of MEPI dimensions by urbanity, 2004 and 2011 …….……......... 45
Figure 7.4. Geographic Mapping of MEPI in the Philippines, 2004 & 2011 ……..…........... 49
Figure 7.5. Contribution of MEPI dimensions by region, 2004 and 2011 ………………….. 46
Figure 7.6. Comparison of MEPI, MPI, income headcount and poverty incidence by region,
2003-2004 and 2011-2012 ……………………..……………………………………………… 52
Figure 7.7. Trend of MEPI in different poverty cut-offs …………………...………...……... 55

ix
An Analysis of the Multidimensional Energy Poverty Index:
The Philippine Case

Alyssa Mari G. Catanjal and Trexia Mae D. Lechoncito

CHAPTER 1: Introduction

What is the extent of the energy poverty in the Philippines?

The energy consumption per capita in the Philippines, has often been cited as low compared

to other South East Asian countries (IEA, 2013) (see Figure 1.1). An index formulated by the

World Economic Forum (WEF) presents the value of energy architecture performance (EAP) per

country that is aligned with energy triangle-economic growth and development, environmental

stability, and energy access and security. WEF’s index shows that among the South East Asian

countries, Philippines has the lowest access to energy and energy security (see Figure 1.2). Despite

efforts from the Department of Energy (DOE) to improve the status of energization in the

Philippines, this problem still persists.

Figure 1.1 Energy consumption per capita in Southeast Asia, 1990-2013


9,000.00
kg of oil equivalent per capita

8,000.00
7,000.00
6,000.00
5,000.00
4,000.00
3,000.00
2,000.00
1,000.00
0.00

1990 2000 2010 2013

Source: World Development Indicators, World Bank


Note: No data for Lao PDR, and 1990 data for Cambodia

1
Figure 1.2 ASEAN’s energy access and security in EAPI (2013-2017)

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0
2013 2014 2015 2016 2017
Brunei Cambodia Indonesia Malaysia
Philippines Singapore Thailand Vietnam

Source: World Economic Forum


Note: No data for Lao PDR and Myanmar

The relationship of energy access to human development has been emphasized by

numerous studies (Kanagawa and Nakata, 2008; Sihag et al., 2004; Pokharel, 2007; Reiche et al.,

2000; and FAO, 2000). Access to energy is one of the most important factors in raising people out

of poverty in developing countries—highlighting the strong correlation between an increase in

energy access to poverty alleviation (Power for all, 2014). Specifically, the benefits of energy

access were further explained as one of the fundamentals in fulfilling basic social needs—paving

the way for improved health and education, better access to information, and enhanced household,

agricultural, and industrial productivity (Gaye, 2008). Likewise, the Food and Agriculture

Organization of the United Nations (FAO) (2000), posits that energy is an essential ingredient for

almost all human activities such as cooking, space/water heating, lighting, food preparation and

storage, and education.

2
With the improvement of health and welfare that energy access brings, more attention has

been paid into improving access to energy in every corner of the world, especially among

developing countries (IEA, 2014). However, despite the known importance that energy access

entails with eradicating poverty and, promoting economic and socio-economic development, there

have only been a few studies conducted in the Philippines that aims to address poverty by

improving energy access, and even studies that quantified energy poverty in the country.

One of the most common approaches that studies apply in quantifying energy access in the

Philippines is a binary approach—a method of identifying whether or not the household has access

to energy—and equating energy access to having access to electricity (ESMAP, 2002 and

Balisacan, 2001). This kind of approach only shows a narrow picture of the problem; failing to

take into account the multidimensional aspect of energy access.

Due to this gap among existing research, our main objective in writing this paper is to look

at the status of energy poverty in the Philippines by including the multidimensional nature of

energy access. We aim to provide insights on the significance of looking at the multidimensional

aspect of energy poverty, and its contribution to the overall poverty in the country. This is to better

identify the energy poor households, which could help in knowing the kind of solutions that will

effectively solve the problem.

3
CHAPTER 2: Trends of Energy Consumption

2.1. Trends of Energy Consumption in the Philippines

Lack of energy access does not only affect Philippine households but also the industries

that highly contribute to economic development. Data released by the Philippine Statistical

Authority (PSA) show that majority of the energy in the Philippines is being consumed by the

transport and the residential sectors (see Figure 2.1). The International Energy Agency (IEA)

(2004) highlights the importance of energy as a prerequisite for economic development, where the

reverse can also be said as true. Economic growth largely influences energy consumption

especially on the early stages of development, because as countries develop and living standards

improve, demand for energy grows much more rapidly (US Energy Information Administration,

2016). As energy access highly contributes to economic development, it is also the case that a

limited access to modern and affordable energy services contributes to the high poverty levels that

developing countries, particularly in sub-Saharan Africa and some parts of Asia, face (IEA, 2004).

This is also the case in the Philippines.

Figure 2.1 Average energy consumption by sector in the Philippines (1998-2015)

manufacturing mining construction


transport residential commercial
agri_fish_forest others

Source: Philippine Statistics Yearbook, 2009-2016

4
Aside from the energy consumption per sector in the Philippines, we explored the trends

in the types of fuel used by households using the final reports of the Household Energy

Consumption Survey (HECS)—a household based survey on usage of different energy sources.

This is important because the type of fuel used by households in the Philippines will be utilized in

the latter portion of this study. We also looked at the trends in the perceptions on the use of

common household energy sources.

As shown in Figure 2.2, electricity consistently displayed the highest proportion of

household usage with about 87.2% at the end of August 2011. The significant increase in electricity

usage from 64.6% to 83.9% in 1995 could be attributed to energy programs employed by the

government to hasten rural industrialization (NSO-DOE). In contrast, biomass residues posted an

all-time low percentage at 22.3% in 2011. This low usage in renewable sources of fuel is in parallel

to the shift in household’s preference to conventional types of fuel in 2004 where a 4.3 percentage

point decrease of charcoal usage and a substantial 10.5 percentage point reduction in biomass

residues usage was posted. Albeit the increase in usage of modern energy sources (electricity, LPG

and kerosene), fuelwood shows a meaningful 54.2% usage at the end of August 2011.

Figure 2.2 Type of fuel used by households in the Philippines, 1989-2011 (in %)

1989 1995 2004 2011


87.6
87.2
83.9

79.9
74.5

67.1
64.7

63.5
56.1

55.1
54.2
52.1

46.4
41.2

38.5

36.4
34.3

34.2
32.1

29.2
33

22.3
21.9

18.7

E LE C TR IC ITY LPG KEROSENE FUELWOOD CHARCOAL B IOM ASS


R E SIDUE S

Source of basic data: 1989 Energy Statistics, 1995, 2004, and 2011 HECS by NSO-DOE

5
In the use of traditional fuels, fuelwood used for cooking leads at 99.6% in 2004 and 99.8%

in 2011 followed consistently by charcoal at 88.1% in 2004 and 97% in 20111. Additionally, there

is an overall 22.5% decrease in the number of households without electricity access from 1989 to

2011. From 1995 to 2004, refrigeration, space cooling, and entertainment and education declined

significantly. Specifically, from 67.4% to 55.1% for refrigeration; from 49.6% to 25.4% for space

cooling, and from 19.4% to 10.2% for entertainment and education. Despite the meaningful

decrease from 1995 to 2004, a minimal increase from 2004 to 2011 is evident.2

2.2. Energy consumption in the Philippines vs. other Southeast Asian countries

An imbalance among the Southeast Asian countries exists when looking at the broader

perspective of energy poverty. Figure 1.1 of Chapter 1 shows the amount of energy consumption

per capita of Southeast Asian countries. Brunei Darussalam and Singapore show the highest

amount of energy consumption with 3,000 kg of oil equivalent (kgOE3) and above. This is

followed by Thailand and Malaysia with an energy consumption between 1,000 and 3,000 kgOE.

And the others with below 1000 kgOE, including the Philippines. However, despite the low energy

consumption per capita in the Philippines and Myanmar, their respective GDP per energy use

tended to increase (see Figure 2.3), highlighting the two countries’ effective use of energy.

Worldwide, an estimate of about 1.3 billion people lack access to electricity, while 2.6

billion still relies on traditional biomass stoves and open fires for cooking (REN21, 2013). This is

also a situation evident in the Southeast Asia, where 120 million lack access to electricity and 276

million people are still reliant on the usage of traditional biomass stoves. However, progress seems

to be evident as there had been a decrease in the number of people within the region who don’t

1
See Appendix Figure A.1.
2
See Appendix Figure A.2.
3
kgOE is the unit used as a common metric to quantify energy supplied using a variety of sources and carriers.

6
have access to electricity by two-thirds (IEA, 2015). Table 2.1 shows the status of electrification

within Southeast Asia, highlighting that the Philippines is third among the region, following

Myanmar and Cambodia, that has the highest number of shares of population (21%) who don’t

have access to electricity. Table 2.2 on the other hand, shows that 50% of the population in the

Philippines uses traditional biomass for cooking.

Figure 2.3 GDP per energy use (constant 2011 PPP $ per kg of oil equivalent)
18
16
14
12
10
8
6
4
2
0
1990 2000 2007 2008 2009 2010 2011 2012 2013 2014
Brunei Cambodia Indonesia
Malaysia Myanmar Philippines
Singapore Thailand Vietnam

Source: World Development Indicators, World Bank


Note: No data for Lao PDR, 1990 data for Cambodia, and 2014 data for Vietnam

Table 2.1 Electricity access in Southeast Asia, 2013


Population without Share of Electrification rate
Countries electricity (in population (%)
millions) (%)
Brunei Darussalam 0 0 100
Cambodia 10 66 34
Indonesia 49 20 80
Lao PDR 1 13 87
Malaysia 0 1 99
Myanmar 36 68 32
Philippines 21 21 79
Singapore 0 0 100
Thailand 1 1 99
Vietnam 3 3 97
Total Southeast Asia 120 19 81
Source: IEA, 2015

7
Table 2.2 Number of people relying on traditional biomass for cooking, 2013
Selected Asian Countries Number (in millions) Share of population (%)
India 772 66
Bangladesh 149 91
Indonesia 128 55
Pakistan 111 64
Philippines 47 50
Vietnam 49 56
All Developing Countries 2588 49
World 2588 38
Source: REN21, 2013

2.3. Perceptions on the Use of Common Household Energy Sources in the Philippines

To analyze the trends in the energy sector in the Philippines, especially the probable

reasons for the shift in the use of renewable energy sources to conventional energy sources, and

vice-versa, non-accessibility to electricity, and non-ownership of a refrigerating unit or

recreational appliances, we summarized the perceptions on the use of common household energy

sources using the HECS final report in 2004 and 2011.

As illustrated in Figure 2.4, the most probable factors considered in the use of common

household energy sources were mostly the prices associated with the installation of each, followed

by convenience. For kerosene and fuelwood, the cleanliness and environmental impact were

accounted for.

8
Figure 2.4 Perceptions of households on the use of common household energy sources, 2004
and 2011

Electricity LPG
Electric appliances are
Other fuels are better…
expensive
The monthly cost of LPG is convenient
electricity is expensive LPG is unsafe
The electricity installation
LPG is expensive
charge is expensive
Applying for an electrcity LPG equipment is…
installation is difficult LPG is hard to get/not…
0.0 50.0 100.0 0.0 50.0 100.0

Kerosene
Fuelwood
Kerosene is unsafe
Kerosene is convenient to… Fuelwood gathering is a…
Kerosene is dirty Cooking with fuelwood…
Kerosene is expensive Fuelwood gives hotter flame
Fuelwood is inconvenient…
Kerosene equipment is…
Fuelwood is dirty
Kerosene is difficult to get
Fulwood is expensive
0.0 50.0 100.0 Fuelwood is difficult to get

0.0 50.0 100.0

Source: Official Report of HECS (2004 & 2011) by NSO-DOE


Note: Category of electric appliances deemed expensive is not available in HECS 2011. And a household
may report more than one use of fuel.

9
CHAPTER 3: Review of Related Literature

3.1. Definition of Energy Poverty

Energy is very important as it has deeply influenced the lives of people all over the world,

most especially the poor. Due to its known importance, over time, numerous efforts have already

been exerted in trying to quantify energy poverty. However, studies which tried to measure energy

poverty solely focused on a household’s usage or consumption of energy devices.

Numerous studies have already tried to define energy poverty. Some define it as the state

of deprivation where the energy consumption of households barely meets the minimum energy

requirements for basic needs (Foster et al., 2000, Modi et al., 2006, OECD and IEA, 2010). The

World Economic Forum (2010) on the other hand, defines it as “the lack of access to sustainable

modern energy services and products.” UNDP (2013) also used the same definition. Others

however have defined an energy poor household as those who are not able to meet their basic

energy needs by estimating the minimum level of energy consumption needed (Parajuli, 2011,

Pereira et al., 2011). Furthermore, others have defined it as the ‘lack of access for modern energy

services’ (Li et al., 2014, IEA, 2014a). Some studies have also tried to quantify energy poverty

by using the expenditure or income parameters, which considers a household to be energy poor if

the household uses 10% of their income on energy (Barnes, 2010). But despite the numerous

amount of studies that have tried to define energy poverty, as of today, there is still no unified

definition for energy poverty (Bersisa, 2016).

3.2. Studies on Energy Access and Energy Poverty


In analysing the context of multidimensional energy poverty in the Philippines, we looked

at the different studies conducted in the country that used energy access as one of the components

of their analysis, as well as studies conducted in other countries that tried to quantify energy

10
poverty. We also explored the literature in the Philippines that used the Alkire-Foster Model,

which is the methodology that is applied in this paper.

3.2.1 Studies that used Energy Access in the Philippines

Table 3.1 displays a tabular summary of studies that quantified and used energy indicators

as one of their independent variables. All the studies that we’ve found uses a binary approach,

which pertains to a single variable that indicates whether or not a household is energy deprived or

not, in measuring energy access

Table 3.1 Summary of Empirical Studies on Quantifying Energy Access in the Philippines
Sources Year Approach Output Results
indicator(s)/Sample
Balisacan 2001 Binary Proportion of the rural Electricity access has a
Approach population living below minimal impact; roads have
the provincial poverty line the greatest impact.

ESMAP 2002 Binary Survey of approx. 28,000 Average incomes of home


Approach domestic, commercial, businesses using electricity
industrial and irrigation are significantly higher than
units with and without those who do not use
electricity electricity but no positive
impact of electrification on
incomes from agriculture
was found.
PIDS 2003 Binary Comparison of Compared to other ASEAN
Approach Philippines’ electrification countries, Philippines’ status
and energy usage rate to of electrification continues
other ASEAN countries to lag behind. This is despite
indicating higher electricity
production growth after the
economic crisis.

11
3.2.2 Cross-country studies on Energy poverty
In almost all corners of the world, numerous studies have already tried to quantify energy

poverty. Table 3.2 shows the different empirical studies among countries that have quantified

energy poverty through various approaches.

Table 3.2 Summary of Empirical Studies on Energy Poverty


Source Year Country Approach Results
Pachauri & 2004 India Two- Energy poverty is
Spreng dimensional/Engi- rapidly decreasing but
neering method inequality in the
distribution of energy
consumption and access
is increasing.
Barnes, et al. 2010 Bangladesh Minimum end-use 58% are energy poor
energy (MEE)
Nussbaumer, et al. 2011 Some MEPI Majority indicates a 0.9
African to 0.6 value of MEPI.
countries
Khandker, et al. 2012 India Demand based 41% are energy poor
despite being income
nonpoor.
Tchereni, et al. 2013 Malawi Energy expenditure More than 80% of the
households are below
the energy poverty line.
Edoumiekumo, et 2013 Nigeria MEPI 83.2% of the population
al. are energy poor.

Sher, et al. 2014 Pakistan MEPI 47% of the population


are energy poor.

Both in Bangladesh and India, energy poverty is measured through a demand approach. In

the demand approach, a minimum level of energy consumption needed to satisfy basic needs is set

where such energy consumption should be insensitive to the income of the household. Barnes, et

al. (2010) defined energy as the useful energy that households consume (end-use energy)4.

4
Actual energy that was made available and/or provided for by total energy used.

12
Khandker, et al. (2012) on the other hand, set an energy poverty line and have set it as the threshold

point where as income increases, energy consumption also increases. Energy poverty in this study

was defined as the level of energy consumption that the household absolutely needed to sustain

welfare.

Pachauri and Spreng (2004) measured energy poverty in India, by combining two different

approaches. First, is by estimating basic energy needs and consumption of households through an

engineering approach and certain normative assumptions. Second, by looking at poverty in relation

to the access of households to different energy sources by computing the energy expenditure in

proportion to its total expenditure. Pachauri and Spreng (2004) adopted a cut-off point of 10% of

total expenditure since it is most commonly mentioned as the level of expenditures for households

that are poor.

Tchereni, et al. (2013) on the other hand, used the expenditure approach to quantify energy

poverty where a household is energy poor if more than 10% of its income is allocated on energy

facilities (Fahmy, 2011). The studies conducted in Nigeria, Pakistan and other African countries

used Multidimensional Energy Poverty Index (MEPI) to measure energy poverty (Edoumiekumo,

et al., 2013; Sher, et al., 2014; Nussbaumer, et al., 2011). MEPI was constructed following Alkire-

Foster’s approach in measuring Multidimensional Poverty Index (MPI) where the headcount (H)

is multiplied to the average intensity of the dimensions identified in the index (A). Further

discussion on MEPI is elaborated in chapter 4 of this paper.

3.2.3 Empirical Studies using Alkire-Foster (AF) Model in the Philippines

In the Philippines, a plethora of studies had been conducted which employed the Alkire-

Foster methodology. However, majority of the studies that we’ve found only used the AF model

in quantifying poverty in a multidimensional nature, showing that was no study yet that used the

13
said model in quantifying energy poverty. Table 3.3 shows a tabular summary of some studies in

the Philippines that adapted the AF methodology.

Table 3.3 Summary of Empirical Studies that used AF Model in the Philippines

Sources Year Data and Topic Results


Sample Period
OPHI 2011 DHS (2008) Multidimensional Compared to other poverty
Poverty Index measures, MPI indicates a lower
(MPI) number of incidence/headcount
are poor in the Philippines.

Casimiro, 2013 DHS (1993- Multidimensional The parameters that the authors
et al. 2008) Approach to used are robust and that under-
Child Poverty five children are deprived in at
least two dimensions.
Balisacan 2015 FIES, NDHS, MPI “Aggregate poverty, seen from
APIS (1998- the lens of multidimensional
2012) deprivation, actually
declined as the economy
expanded during the past
decade.”
Datt 2016 APIS (2004- Multidimensional There is a significant decline in
2013) Poverty multidimensional poverty in the
Philippines compared to official
measures of poverty.
Albis and 2017 FIES-LF (2012), Employment Employment, the number of
Elviña CPH Barangay Correlates young and old dependents,
Schedule (2010 community public goods
contributes to MPI

3.3. The Role of Energy Access

In many of today’s global development challenges like poverty, inequality, climate change,

food security and even health and education, energy has always been a part of the central solution

in addressing such concerns (Nussbaumer, et al., 2011). Thus, one cannot deny the high importance

of energy into numerous different aspects- may it be for economic growth, poverty, human

development, health and education, and gender equality.

14
Impacts on Poverty and Household Income

Energy is recognised as one of the most essential inputs in sustaining people’s livelihood

which has led to new approaches that tried to include energy as a means for poverty alleviation

(Clancy, 2003). This theory paved way for several empirical studies that have tried to quantify if

whether an increase in electricity access could benefit the poor and help in increasing their income.

Fan et al. (2002) found in their study in China, that a positive relation between the amounts of

infrastructure investment in rural electrification and the number of poor reduced for every 10, 000

yuans invested in infrastructure exists. Fan et al. (2004) discovered a similar result in Thailand,

where among all the different public investments that they have identified, rural electrification has

the largest poverty reduction impacts. IEA (2004) adds that a limited access to modern and

affordable energy services much likely contributes to the high poverty levels that developing

countries, particularly in sub-Saharan Africa and some parts of Asia, face which can also be said

with the case of the Philippines.

Various empirical studies find that increasing electricity access contributes to poverty

reduction. However, despite this, some studies find otherwise. Fan et al. (1999) claims that an

additional government spending on rural electrification, has no discernible impact on decreasing

poverty in India. Balisacan (2001) also discovered the same thing, highlighting that electricity

access, although significant, it barely has an effect on reducing the proportion of the rural

population living below the official poverty line of the country.

In a micro perspective, ADB (2005) conducted a study in China, India and Thailand about

the impact of electricity access to the income of poor households. They found a negative

correlation exists between the degree of electricity access and income of poor households in

Thailand, and that a higher income inequality exists in electrified villages than in unelectrified

15
villages in India. They also discovered that in China, households in Shaanxi has a faster income

growth than the non-electrified poor households. However, they did not discover electricity to have

a positive impact on poverty level.

ADB (2010) concluded in their study in Bhutan that electricity access has a significant

effect on nonfarm income. ESMAP (2002) also found a similar finding in the Philippines, stating

that electrification has no positive impact on incomes sourced from agriculture. Khandker (2009a)

however, contradicted this result in her study in Vietnam where she concluded that grid

electrification has a strong influential increase in farm income and has hardly no effect for non-

farm income. Khandker (2009b) further supported Khandker’s (2009a) results in his study in

Bangladesh, highlighting that electricity access has a positive impact on both farm and non-farm

incomes. She adds that electricity’s positive effect on income is four times higher for wealthier

households than for poorer households.

Fan et al. (2005) overall, concluded that an access to electricity in Tanzania, significantly

increases household income in all economic zones. They approximated that a 1% increase in the

electrification rate would approximately lift 140,000 people out of poverty. Khandker (2009a)

further supported this claim, stating that grid electrification increases household income by at least

25%. UNDP (2011) also mentions that electricity access explains 30% of the increase in household

income in Nepal.

Impacts on Human Development

Energy is considered as an engine for socio-economic development (Nussbaumer, et al.

2013). It is believed that the standard of living and energy consumption are positively related with

one another, and the higher the standard of living for a certain country is, the higher the energy

16
consumption (Wu et al., 2010). However, there are only a few literatures that discuss the

relationship between human development index (HDI) and energy consumption.

Martinez and Ebenhack (2008) studied the correlation of HDI and energy consumption per

capita for 120 countries. They found a strong relationship between the index and energy

consumption for majority of the countries. They were also able to identify three important trends

based on the data: a steep rise in human development relative to energy consumption for energy-

poor nations; a moderate rise for transitioning nations; and no rise for developed nations

consuming large amounts of modern energy (Wu et al., 2010). Pasternak (2000) found a similar

finding using the 1997 data where he concluded that HDI and energy consumption per capita are

highly correlated with each other.

Ediger and Tathdil (2007) also supported the same findings when they recalculated the

weights of the 2000 HDI data for 173 countries by reintegrating energy as a component of the

index. They discovered severe differences after comparing the modified index to other energy-

related indicators provided in the UN Human Development Report of 2003.

17
CHAPTER 4: Existing Energy poverty metrics

Measuring energy poverty has started to become a focus among researchers. However, a

unified approach on how to quantify energy poverty is still inexistent. In the following, we present

different methods that are currently being used.

Binary Metrics Approach

The binary metrics approach highly focuses on a single dimension relying on only one

threshold of energy supply to assess if whether a household can be considered as energy poor. This

is measured more commonly by finding out whether a household have access to energy or not.

And although this kind of approach is easy to collect data for and interpret, this methodology only

presents a very narrow picture of energy access among households-failing to take into account the

multidimensional aspect of energy access (ESMAP, 2015).

Minimum Energy Consumption Threshold Approach/Energy Poverty Line

The concept of a poverty line has been one of the most relied on approach all over the

world. Whereby the methods of defining a poverty line are often based on an expenditure approach

where the minimum level of expenditure is set at which household can be considered as non-poor.

And likewise, this is also the same approach as to how energy poverty line is derived (Barnes et

al., 2010). An energy poverty line is defined as the threshold to quantify how much energy

consumption is needed to maintain a bare minimum livelihood for households (Barnes et al., 2010).

An approach in measuring energy poverty was adopted by Bravo et al. (1979) where he

considered the different variations in energy sources and their efficiencies, climate conditions, and,

urban and rural areas to quantify a household’s direct energy5 needs. Bravo et al., (1979) defined

5
Direct energy is defined as the usage of energy for cooking, lighting, heating/cooling, preservation of food, hot water,
ironing, pumping water, recreation and other social occasions. (Barnes, et al., 2005)

18
energy poverty line as the minimum quantity of energy needed by households to have a reasonable

quality of life. The method employed by Bravo et al., (1979) defined the average essential

household level for direct energy needs as 27.4 kgOE6 per capita per month.

Modi et al., (2005) defined energy poverty as the inability of the household to cook using

modern cooking fuels, as well as the inability to have a bare minimum amount of electric lighting

to do household and other productive activities. They have set the minimum amount needed to 50

kgOE7 per capita per year. This is based on the absolute minimum requirement of 40 kgOE for

cooking and 10kgOE for lighting. However, this kind of measure do not consider the energy

consumption used for agriculture, transport and other community-level needs like social services

and government activities (Modi et al., 2005). This kind of approach also fails to account market

conditions like prices and other energy policies that affect the delivery of energy services (Barnes

et al, 2010). Sanchez (2010) on the other hand, proposes a threshold level of 120 kilowatt-hours

(kWh) of electricity and the equivalent of 35 kilograms (kg) of LPG per capita per year.

Energy Expenditure Approach

A different approach in measuring energy poverty, as adopted by Pachauri and Spreng

(2004), used energy expenditure in proportion to the total expenditure of a household. The said

approach is applied as such because among different expenditure surveys, poor households tend

to use much of their income with expenses related to energy usage. So an increase in the

household’s total expenditure leads to a decrease in the household’s energy expenditure up until

they decrease it into a minimum level of 10% set as cut-off point by Pachauri and Spreng (2004).

6
Bravo et al., (1997) stated it as 9.2 thousand kcal/day/person which is then converted to as 27.4 kgOE per capita per
month.
7
kgOE is the unit used as a common metric to quantify energy supplied using a variety of sources and carriers.

19
Relatively, this kind of method has been considered to be easily administered in studies especially

in household surveys, as the official income or expenditure is readily available. However, this

approach assumes that income-poor or expenditure-poor are energy poor as well. But this is not

always the case (ADB, 2015).

Aside from the method applied by Pachauri and Spreng (2004), another study has also

applied the energy expenditure approach as a means to quantify energy poverty. Foster, et al.

(2000), have set an energy poverty line where those households that have a per capita energy

expenditure that is exactly or below the overall expenditure poverty line estimated for the country,

are the ones considered energy poor.

Energy Development Index (EDI)

Constructed by IEA in 2004, the primary aim of the Energy Development Index (EDI) is

to track a country’s progress in transitioning from the usage of traditional to modern fuel. The

index is measured through national data, composed of four indicators—commercial energy

consumption per capita, electricity consumption in the residential sector per capita, share of

modern fuels in total residential sector energy use, and share of population with access to electricity

(IEA, 2010). As the composite index makes use of a national data, EDI does not provide

disaggregated information about the degree of energy access, factors that impede access to energy,

and who are those mostly affected by the energy deficiency (ESMAP, 2015).

Multidimensional Energy Poverty Index (MEPI)

Efforts in trying to quantify energy access have started to gain attention especially since

the call of Sustainable Energy for All (SE4ALL) to achieve “universal access to modern energy

access by 2030.” However, quantifying energy access was faced with complications as there have

20
been no universal definition of energy access as well as measuring such definition in an accurate

manner (World Bank Group, 2014).

Initially, energy access was measured in a binary metric (‘access or no access’) approach

where a household is measured to have energy access if they have electricity and energy poor if

they don’t. And if whether they are cooking with solid or non-solid fuels. But this kind of approach

of measuring energy poverty is rather limited as it does not take into account the relevance of

energy access to industries and other community sectors like schools and hospitals.

Bourguignon and Chakravarty’s (2003) argument about poverty is that a single

dimensional approach defines poverty insensitively—failing to take in to account each dimension

of an individual’s well-being.8 This could also be applied in the context of energy poverty. This

drives the reason as to why Alkire and Foster (2009) argued that poverty should be looked at the

different dimension-specific cut-offs. However, due to the limitation that using dimensions-

specific cut-offs alone entails-failing to take a look across dimensions of poverty- Alkire and

Foster introduced a model that would try to aid in solving this problem.

Before, the methodology in trying to quantify energy poverty is through the unidimensional

approach where a minimum energy consumption threshold is set. If the energy consumption of a

household falls below this threshold, they are then considered as poor. From the perspective of the

capability approach, viewing multidimensional energy poverty through the lens of unidimensional

would then fail to take a look at relevant information of important dimension-specific shortfalls

(Alkire and Foster, 2009).

As the importance of identification method that focuses on dimensional shortfalls is

highlighted, as argued above, different criterions then popped out as to how to identify the poor.

8
See also Chakravarty et al 1998 and Tsui 2002 on this point.

21
One criterion that uses this type of identification is the union method of identification. This

identification often predicts a very high number result. This is because in this method, among all

the different dimensions of poverty, an individual is identified to be poor if s/he is deprived in any

one of the different dimensions (ci>1) (Alkire and Foster, 2009). As the number of dimension

increases, the union approach will often identify that majority of the population is poor, even if

some person that was said to be poor are not really considered as one. Some examples of literature

that used this approach are Charavarty, et al. (1998), Tsui (2002), Bourguignon & Chakravarty

(2003).

Another multidimensional identification method is the intersection approach. Opposite to

the union approach, the intersection approach identifies the poor if s/he is deprived in all of the

different dimensions of poverty (ci=d) (Alkire and Foster, 2009). And rather successfully, this

approach would accurately be able to identify the poor especially if the dimensions identified are

sufficient. And truthfully, in contrast to the union approach, this identifies a rather small amount

of number, focusing on the poorest of the poor. But this approach also inevitably misses out

identifying some people “who are experiencing extensive, but not universal, deprivation” (Alkire

and Foster, 2009).

Due to the limitations of the union and intersection approach, Alkire and Foster tries to

create a cut-off that could identify more accurately the poor, the dual-cutoff method. This method

uses an intermediate cut-off level for ci that lies in in between the extreme values of 1 and d. The

person identified as poor is then defined as those people whose number of deprivation falls above

or equal the cut-off k (ci≥k). And this is also the same approach that was employed by Nussbaumer,

et al. (2011) in identifying the multidimensionally poor.

22
The dual-cutoff measure entails characteristics that is one, ‘poverty focused’, in a sense

that an increase in the achievement level of an individual that is not poor leaves it unchanged. And

secondly, it is also ‘deprivation focused’ which means that a person’s poverty status is independent

of the changes in the levels of non-deprived achievements (Alkire and Foster, 2009).

The multidimensional approach using Multidimensional Energy Poverty Index (MEPI)

was developed by the Oxford Poverty and Human Development Initiative (OPHI) using the

foundations that was laid out by Alkire-Foster’s method of measuring Multidimensional Poverty

Index (MPI) (Nussbaumer, et al., 2011). Through the multidimensional approach, “access is

defined as the ability to obtain energy that is adequate, available when needed, reliable, of good

quality, affordable, legal, convenient, healthy, and safe for all required energy.” (World Bank

Group, 2014).

Fundamentally, MEPI “captures the set of energy deprivations that may affect a person”

(Nussbaumer, et al., 2011) whereby a person is considered to be energy poor if the combination of

the different deprivations experienced by the person exceeds a pre-defined threshold. MEPI

consists of five energy dimensions (i.e. cooking, lighting, appliances/services, communication,

education and entertainment) to represent basic energy needs using six indicators. In summary,

MEPI is the product of a headcount ratio (H) (share of people identified as energy poor) and the

average intensity of deprivation of the energy poor (A) (Nussbaumer, et al., 2011).

As used by OPHI, they have identified the different dimensions and indicators for MEPI

which are enumerated in Table 4.1.

23
Table 4.1 Dimensions, Indicators and Variables used for MEPI

Dimensions Indicators Weights Variables


Cooking Modern cooking fuel 0.20 Type of cooking fuel

Indoor pollution 0.20 Food cooked on stove


or open fire
Lighting Electricity access 0.20 Has access to
electricity
Household Appliances Household appliance 0.13 Has a fridge
ownership
Entertainment/education Entertainment/education 0.13 Has a radio or
appliance ownership television
Communication Telecommunication 0.13 Has a mobile phone or
means a land line phone

Source: Nussbaumer, et al., 2011 & 2013

24
CHAPTER 5: Conceptual Framework

People are living in poverty when the well-being that they experience is below the

minimum acceptable level (Schaffner, 2014). One measure in identifying the poor is through the

usage of a poverty line that represents the value of well-being indicators that marks the minimum

acceptable level.

Energy Poverty Line

Official poverty lines among different countries varies tremendously. Majority of the

developing countries define their poverty line in absolute terms, but vary as to how they define the

identified minimum acceptable levels of acceptable goods. Some developing countries, on the

other hand, define poverty lines in relative terms (Schaffner, 2014).

Just as how poverty lines are constructed and defined, we adopted Schaffner’s framework

to define an energy poverty line. People are identified as living in an absolute energy poverty

situation if they lack access to modern energy services like clean cooking fuels, modern stoves and

are deprived of using a reliable and efficient lighting, and even telecommunication services (IEA

et al., 2010). In contrast, people are identified as relatively energy poor if their access to modern

energy services are low compared to the typical level in their society (Schaffner, 2014).

Figure 5.1 illustrates how concepts like energy consumption gap, proportional energy

consumption gap, headcount ratio and so forth are determined through the energy poverty line.

The said figure also illustrates how energy poverty is determined through a minimum energy

consumption threshold. The figure shows energy consumption per capita in a person’s household

(Y), where the N individuals in a population is ordered from 1 to N with 1 being the poorest and N

being the richest. The person’s rank order (i) is used in the horizontal axis alongside with their

energy consumption (Yi) in the vertical axis. The average height of the figure represents the average

25
level of energy consumption per capita, while the slope is a representation of how energy

consumption rapidly rises as you move from the poorest to the richest individual.

Figure 5.1. Energy poverty line


Energy consumption (Y)

Source: Constructed by the authors as adopted from Schaffner (2014)

z in Figure 5.1 represents the minimum acceptable level of energy consumption per capita

where z touches the Y schedule at person index q, which represents that of the N number of

population, there are q people who are energy poor. The vertical distance of z-Yi represents person

i’s energy consumption gap where if divided to z, as in (z-Yi)/z, represents person i’s

proportional energy consumption gap. If this value is substituted to 0.30, it represents that the

person’s energy consumption falls 30% below the energy poverty line. The equation q/n on the

other hand represents the number of households in population whose energy consumption falls

below the energy poverty line or most commonly called as the headcount ratio. The shaded part

in Figure 5.1 represents the total amount of energy consumption needed to bring all energy poor

people above the energy poverty line or the total energy consumption gap.

26
Figure 5.2 on the other hand, outlines how overall energy access index is measured by

considering energy access among households, within production and even in the community.

However, for this study, we would mostly be focused on the construction of the index of

household’s access to energy since the unit of observation of our paper, is of the household level.

This will provide the foundation on the indicators that will be used in the construction of the

multidimensional energy poverty index or MEPI as deemed applicable to the context of the

Philippines and based on the variables available in the data of HECS. Aside from this, the number

of dimensions will serve as a way to identify properly the headcount ratio—a concept defined

previously in the section about energy poverty line—of those who are multidimensionally poor.

Figure 5.2. Energy access multi-tier framework

Source: ESMAP’s A New Multi-Tier Approach to Measuring Energy Access (2014)

Energy availability in reducing poverty

Figure 5.3. emphasizes the importance of adequate energy availability to poverty reduction,

where an increase in the energy availability could lead to an increase in energy consumption ceteris

paribus. This figure further highlights that an increase in energy access contributes to poverty

alleviation, which then increases welfare. Such an increase in energy consumption leads to changes
27
in socio-economic indicators like employment, and productivity that could then increase income

and GDP that would on the latter reduce poverty. However, it is important to note that energy

availability and energy consumption may be endogenous with one another, and thus requires

attention when being applied to the data.

Figure 5.3. Steps from energy availability to poverty reduction

Energy Availability (adequate


quality and quantity)

Energy consumption/use;
adoption of appliances

• Changes in enterprise/use of appliances; new


products/services; changes in quality,
productivity, cost of production, prices or
volumes of product, operating hours,
employment
• Creation of new enterprises

Changes in income, GDP


and other benefits

Poverty reduction

Source: From ESMAP’s The Impact of Electricity Access on Economic Development: A Literature Review
as adopted from Koojiman-van Dijk (2008)

28
CHAPTER 6: Data and Empirical Approach

6.1. Data

The primary source of data used in this study comes from the Household Energy

Consumption Survey (HECS), a household survey jointly conducted by the Department of Energy

(DOE) and Philippine Statistics Authority (PSA). HECS collects data on households’ usage of

different energy sources such as LPG, kerosene, etc., and where these sources are being used for

(i.e. cooking, lighting, ironing, etc.). The data set for 1989 and 1995 HECS were not available,

thus, the researchers only utilized the 2004 and 2011 data. HECS derived its sampling design from

the 2003 Master Sample (MS) for household-based surveys. It only used half of the households

(about 25,000 households) which are deemed to be sufficient to quantify the trends in energy

consumption in the national level. Thus, the sample households serve as a national representative

of the whole population with the application of the respective sampling weights of each year. The

total sample households for HECS 2004 is 21,960 while HECS 2011 has 20,591. Using a survey

data design, there were approximately 17.0 million households in the country in 2004 while, there

are about 21.0 million households in 2011. The energy sources mostly utilized in this study are

categorized to modern and traditional energy sources. With electricity, LPG and kerosene

comprising modern energy sources and charcoal, firewood, and biomass residues as traditional

energy sources.

The Family Income and Expenditure Survey (FIES) for the years 2003 and 2012 was also

used for constructing the multidimensional poverty index (MPI) which will be used for further

analysis of the results of MEPI. FIES is a household based nationwide survey conducted by the

National Statistics Office once every three years. The total sample households for FIES 2003 is

42,094 while FIES 2012 has 40,171.

29
6.2. Summary Statistics

Select Household Characteristics

Table 6.1 and 6.2 show select household characteristics only for the year 2011 due to

limitation of data in HECS 2004. Among the households that participated in the survey, 80.11%

are male-headed households, with an average age of 50.5 and an average family size of 5.

Table 6.1 Summary Statistics of Select Household Characteristics (n = 20, 591)

Variable Mean Std. Dev. Min Max


Male-headed HH (in %) 80.1 - 0 1
Female-headed HH (in %) 19.9 - 0 1
Age 50.534 13.892 15 98
Total Number of Household
Members 4.817 2.269 1 20
Source of basic data: HECS 2011

As in the case of income, both in cash and in kind, it is shown in Table 6.2 that in both

2004 and 2011, an average of 68% and 64%, respectively, of the respondents surveyed belong to

households with an income less than 10, 000. This group also consists majority of the respondents

who answered the survey followed by households with 10,000 to 29,999 income at about 26% in

2004 and 27% in 2011.

Table 6.2 Summary Statistics of Household’s Average Monthly Income (in cash and in
kind; 2004 and 2011)

Variable Mean Std. Dev


2004 2011 2004 2011
Less than 10,000 .681 .638 .466 .482
10,000 to 29,999 .258 .272 .438 .445
30,000 to 59,999 .047 .072 .212 .258
60,000 to 99,999 .010 .014 .099 .116
100,000 and over .005 .005 .067 .068
Source of basic data: HECS 2004 & 2011

30
As for the different dimensions of energy access we used, Table 6.3 displays an average of

14% of respondents do not have access to electricity both in 2004 and 2011. This is also the case

for indoor pollution, household appliance, space cooling, and entertainment and education where

they all showed a decrease in terms of the average percentage of household who do not have access

to these dimensions. In contrast, the percentage of household who uses traditional cooking fuels

increased from 45.6% to 55.8%.

Table 6.3 Energy dimensions, 2004 and 2011


Variable Name Variable Mean Std. Dev.

2004 2011 2004 2011


Modern Cooking modern_cook 0.456 0.558 0.498 0.497
Fuel
Indoor air pollution 0.741 0.435 0.438 0.496
pollution
Electricity light_elec 0.141 0.142 0.348 0.349
Access
Household ref 0.670 0.650 0.470 0.477
Appliance
Ownership

Space Cooling space_cool 0.470 0.410 0.499 0.492


Entertainment ent_own 0 .282 0.246 0.450 0.431
and Educational
Asset Ownership
Source of basic data: HECS 2004 and 2011

6.3. Empirical Approach

We employed the methodology of a Multidimensional Poverty Index (MPI) of Balisacan

(2015) and the Multidimensional Energy Poverty Index (MEPI) methodology of Bersisa (2016)

and Nussbaumer et.al (2011). Their methodologies were derived from the Oxford Poverty and

31
Human Development Initiative (OPHI) 9 as influenced by Amartya Sen’s argument “for the need

to focus on human poverty by considering the absence of opportunities and choices for living a

basic human life” (Nussbaumer, et al. 2011).

In the methodological approach of measuring multidimensional poverty, Balisacan (2015)

summarizes that the evaluation of multidimensional poverty pursues the same process as the

unidimensional approach. He elaborated that both methods involve defining appropriate indicators

that characterize the dimensions of deprivation, setting deprivation threshold/cut-offs linked to the

identified dimensions, and aggregating the individual information on the poor into a

comprehensive summary measure of poverty. In addition, he considers that a good

multidimensional poverty measurement must come with properties of decomposability and

flexibility. These attributes allow a further disaggregation of the index by subpopulation groups

(region, urbanity, sex, etc.) or by source of deprivation and utilization of “various types of

household survey data, particularly data involving a mix of ordinal (or categorical) and cardinal

indicators of household welfare” (Balisacan 2015).

This approach has been modified to fit the objectives of particular studies that used Alkire

and Foster’s (2011a) method in quantifying poverty.10 As for Bersisa (2016) and Nussbaumer et.al

(2011), a quantification of energy poverty was pursued to aid in policy creation and a more

efficient targeting of the poor to eradicate poverty.

Our analysis of energy poverty is based on the computed MEPI using the Alkire-Foster

Methodology utilized by the OPHI.

9
See works of Alkire and Foster (2007) and Alkire and Santos (2010)
10
See Casimiro, et al. (2013), Edoumiekumo, et al. (2013) and Nussbaumer, et al. (2012)

32
6.3.1 Multidimensional Energy Poverty Index (MEPI)

MEPI is now becoming one of the most common approaches used by researchers to

quantify energy poverty while at the same time taking into account its multidimensional nature.

MEPI considers both the incidence and intensity of energy poverty among households. It follows

the OPHI methodology using the identification and aggregation approaches of Alkire and Foster

(2011a) in constructing the MPI.

Balisacan’s (2015) methodology of MPI and Bersisa (2016) and Nussbaumer et al.’s (2011)

methodology of MEPI are applied in this analysis. However, due to the limitations in our data,

telecommunication means as measured by ownership of a landline or mobile phone are not

considered. Nonetheless, space cooling as measured by ownership of an electric fan/air

conditioning unit is added with accordance to ESMAP’s energy access framework. However,

unlike in Fig. 5.2. of the conceptual framework, the dimension for water heating is not included

for its insignificance in the Philippine context. Given the flexibility of the AF methodology, the

dimensions, indicators, weights, and deprivation cut-offs are selected in accordance with the

objectives of quantifying energy poverty. These are provided in a tabular summary in Table 6.4.

33
Table 6.4 Description of dimensions, weights, and their cut-off points for computing MEPI

Dimension Indicator Variables Weights Deprivation Cut-off


(Poor if…)

Modern Type of cooking 0.20 Household uses


cooking fuel fuel traditional source/s of
energy for cooking
(fuelwood, charcoal)
Cooking Indoor air Type of cooking 0.20 Household uses non-
pollution appliance efficient stove or
appliances such as gas
range. rice cooker,
pressure cooker
Lighting Electricity Has access to 0.20 Household does not
access electricity have access to
electricity

Services Household Has a fridge 0.13 Household does not


provided appliance have a fridge
by means ownership
of Has an electric 0.13 Household does not
household fan or air- have an electric fan or
appliances conditioning air-conditioning
appliances appliance.
Entertainment & Entertainment Has a radio, tap, 0.13 Household has none of
Education or educational TV or computer these assets
appliance
ownership

The index is modified depending on the limitations of the data in HECS, years 2004 and

2011. The index of multidimensional energy poverty is composed of four dimensions, with six

corresponding indicators which include “commonly demanded household energy services”

(Nussbaumer et al. 2011).

Since cooking is a vital activity for the sustenance of life, we included the type of fuel used

by a household in cooking to be able to capture the elements of energy poverty in the use of

traditional fuels such as fuelwood, charcoal, and biomass residues that accord with opportunity

34
cost as compared to modern forms of fuel like LPG, kerosene, and electricity. This is shown in

some studies as majority of women and children use most of their time for daily chores particularly

in collecting fuel for cooking thus losing valuable time for leisure and other meaningful activities

(Nussbaumer et al. 2011; Bersisa 2016).

In the Philippines, some households in the regions still practice the use of traditional forms

of fuel in cooking. As reported in HECS 2011 final report, about 54% use fuelwood as a source of

fuel for cooking while placing third and fourth is the use of charcoal and biomass residues at 35.3%

and 20.1%, respectively.

Additionally, under cooking, indoor air pollution was added to account for households that

use non-efficient cooking appliances. Evidently, females experience direct exposure to this

pollution as they are perceived to do most of the household chores particularly cooking. Thus, we

want to consider the impact of indoor air pollution to health as a significant contributor to energy

poverty in the households.

We also added the space cooling dimension which is in contrast with how Nussbaumer, et

al. (2011) composed their indicators when they constructed the Multidimensional Energy Poverty

Index (MEPI). We added this dimension due to its relevance to the tropic characteristic of the

Philippines.

In accordance with Nussbaumer, et al., (2011) selection of energy access indicators, we

adapted other indicators such as access to electricity and ownership of appliances. Notably,

electricity provides numerous developmental benefits, and is most commonly used for lighting.

Additionally, other services such as entertainment and education are all added to the dimensions

of MEPI as they are also contingent with the usage of electricity. Hence, we added indicators

related to appliances such as fridge, television, radio and computer/printer to be able to take into

35
account the end-use aspect, as it is most commonly not included in energy access metrics. With

the usage of the aforementioned indicators, the element of affordability where the financial means

of households to pay for fuel and other appliances, as well as their benefits to socio-economic

development, are all being taken into consideration.

Applying Nussbaumer and his colleagues’ (2011) methodology, multidimensional energy

poverty was quantified in six indicators of regularly utilized household energy services. The vector

Y={yij} depicts the n×d matrix of achievements for I households across j indicators., thus a value

of yij>0 represents household i’s achievement in the indicator j. We also specify a vector zj that

represents the deprivation cut-offs which are conditions to be met in indicator j such as if yij = 1

then a household is considered not deprived. Then, a column vector denoted by ci is defined as the

weighted deprivation count of a person. Additionally, a poverty cut-off denoted by k is set such as

if ci > k then a household is censored as non- deprived (Nussbaumer et al., 2011).

As for the weights in the fourth column of Table 6.4, we adapted the weights applied by

Nussbaumer et al. (2011 & 2013) to put greater emphasis on cooking, indoor air pollution, and

electricity access. These indicators are shown to be significant energy services as compared to

household appliance ownership and ownership of entertainment or educational devices in the

Philippines. The A-F methodology is divided into two: identification and aggregation.

The identification phase uses a dual cut-off approach which is comprised of the deprivation

cut-off and poverty cut-off. The deprivation cut-off denoted by zj is the minimum acceptable level

of energy consumption as adapted in Schaffner (2014)’s conceptual framework which determines

the energy poor people and provides an estimate of the headcount ratio and total energy access

gap. The cut-offs were determined as per the availability of a certain variable to a household and

36
the poverty cut-off denoted by k used was 0.30 as used by Balisacan (2015) and Nussbaumer et al.

(2011).

The aggregation phase involves the computation of the MEPI as the product of the energy

poverty headcount ratio and the energy poverty intensity (Alkire and Foster, 2011).

MEPI as per Alkire-Foster Model is represented as:

MEPI = H x A (1)

H = incidence/ headcount of multidimensional energy poverty

A = intensity of multidimensional energy poverty

Decomposing the MEPI we have the energy poverty headcount (H) and the energy poverty

intensity (A). As shown in equation (2), the energy poverty headcount accounts for the average

headcount ratio for a household that has a weighted deprivation count (ci) greater than 0.3 (k).

Energy Poverty Headcount:


𝑞
1
H = ∑ 𝑐𝑖 > 𝑘 (2)
𝑁
𝑖=1

k = energy poverty line

ci = number of people who are energy poor or has a deprivation score higher than the cut-

off point/ energy poverty line (k)

N = total number of households

37
As shown in equation (3), the energy poverty intensity is the depth of energy poverty

experienced by a household. It is measured as the average of the censored weighted deprivation

counts with a denominator of the total number of people who are energy poor.

Energy Poverty Intensity:

∑𝑞𝑖=1 𝑐𝑖 (𝑘) (3)


A= ⁄ 𝑞
∑𝑖=1 𝑐𝑖

∑𝑞𝑖=1 𝑐𝑖 (𝑘) = average of the censored weighted deprivation counts

∑𝑞𝑖=1 𝑐𝑖 = total number of people who are energy poor

38
CHAPTER 7: Trends, Analysis, and Discussion of Results

7.1. Trends of MEPI Indicators

The indicators used in the construction of the MEPI were analysed with the aid of HECS

2004 and 2011. We explore the trends by geographical location and sectors to observe

developments in terms of their access to electricity, cooking fuel used, ownership of a refrigerating

unit and ownership of educational and entertainment appliances.

7.1.1. By Urbanity

As shown in Figure 7.1. households located in rural areas are consistently energy deprived

in all dimensions with about 64% and 70% in 2004 and 2011 under the use of modern cooking

fuels. Households in urban areas posted a lower percentage of households who don’t have access

to electricity at roughly 5% in 2004 and 6% in 2011 compared to those in the rural areas where at

about 22% and 20% of the households surveyed are deprived of electricity.

Overall, there is a trend of decreasing number of households without access to electricity,

without space cooling appliances, refrigerator, and no ownership of educational and entertainment

appliances. Meanwhile, the share of households using traditional cooking fuels increased from

2004 to 2011 for both areas.

39
Figure 7.1. Households without access to each MEPI indicators by urbanity, 2004 and 2011
(in %)

100
Urban Rural
80
Percentage

60

40

20

0
2004 2011 2004 2011
Cooking Lighting Refrigerating unit Space Cooling Entertainment and Education
Source of basic data: HECS 2004 and 2011

7.1.2. By Region

As shown in Figure 7.2., ARMM consistently displayed the highest percentage value of

deprived households in all indicators where almost 95% do not have a refrigerating unit in their

homes in 2004. This is followed by Zamboanga del Sur and CAR, with roughly 74% of households

deprived of traditional cooking fuels, and 81% are without space cooling devices in 2004 for CAR.

Meanwhile, NCR posted the lowest percentage of energy deprivation in all indicators. Followed

by CALABARZON and Central Luzon in most of the indicators with approximately 6% and 5%

in no access to electricity in 2004, correspondingly. In totality, there is an average decreasing trend

from 2004 to 2011 in households without access to electricity, no space cooling appliances,

refrigerator, and no ownership of educational and entertainment appliances. On the other hand, the

use of traditional cooking fuels such as fuelwood and charcoal showed an increasing trend.

40
Figure 7.2. Households without access to each MEPI indicators by region, 2004 and 2011

2004
100
90
80
70
Percentage

60
50
40
30
20
10
0
I II III IVA IVB V VI VII VIII IX X XI XII NCR CAR ARMM XIII

2011
100
90
80
70
Percentage

60
50
40
30
20
10
0
I II III IVA IVB V VI VII VIII IX X XI XII NCR CAR ARMM XIII

Cooking Lighting Refrigerating unit Space Cooling Entertainment and Education


Source of basic data: HECS 2004 and 2011

41
7.1.3. By Industry Sector and Gender

The agriculture and fishery sectors are the most deprived in all the indicators with the use

of traditional fuels in cooking yielding the most percentage points. Meanwhile, the finance sector

displays the lowest percentage at 0% with no access to electricity (see Appendix Figure A.3.).

By gender of the household head, the male population posts significant values for all the

indicators vis-a-vis the female population. This trend could be characterized by the significant

number of male headed households with 80.11% vis-a-vis 19.89% female headed households.

Furthermore, access to electricity, entertainment, and education displays minimal discrepancy

between the male and the female whereas services, cooking, and space cooling have a gap of more

than 6% points (see Appendix Figure A.4.).

7.2. Results and Discussions

In this section, the results and analysis for the computed indices H, A, and MEPI are

presented. Furthermore, spatial statistical analysis was implemented to aid in displaying the

different characteristics of MEPI by region.

Multidimensional Energy Poverty Index Results

Table 7.1 summarizes the results of MEPI in 2004 and 2011, and its components--

headcount ratio and average intensity. It highlights a decrease in MEPI from 0.385 in 2004 to 0.327

in 2011 which means that 32.7% of the population is energy deprived of the total potential

deprivations possible in 2011. This decline was caused mainly by the large fall in the headcount

and intensity ratios which is reflective of the efforts of the government to increase rural

electrification rate and provide a stable and reliable power supply among regions 11. The average

11
See Philippine Energy Plans of DOE and Philippine Development Plans of NEDA

42
energy poor household is energy deprived in 63.0% of the weighted indicators in 2004 and 59.3%

in 2011.

Table 7.1 MEPI and its components H and A

Multidimensional Energy Average Intensity of Multidimensional


HECS Poverty Headcount Ratio Multidimensional Energy Poverty Index
(H) Energy Poverty (A) (MEPI)
2004 0.611 0.630 0.385
2011 0.552 0.593 0.327
Note: Poverty cut-off is k=30%

Unpacking the index, we are able to decompose the relative contribution of each dimension

to overall MEPI. Table 7.2 displays that cooking and services provided by household appliances

contribute the most to energy poverty with 48.52% and 34.73% in 2011, respectively. This was

followed by lighting, entertainment, and education with minimal contributions at 8.49% in 2004

and 8.91% in 2011, which could be attributed to the increased efforts of the government to extend

the electrification rate of the country.

Table 7.2 Breakdown of MEPI into its dimensions

Dimension and Contribution to MEPI (%)


Service
provided
Entertainment
HECS Cooking Lighting by means of Total
and Education
Household
Appliances
2004 51.73 6.43 33.35 8.49 100.00
2011 48.52 7.84 34.73 8.91 100.00

We further employ the decomposability feature of MEPI to show the distribution by those

who are considered MEPI poor in terms of geographic area (region & urbanity). We provide a

43
comprehensive profile of the population, headcount ratio, average intensity, and MEPI across

geographic areas for the years 2004 and 2011.

Amidst, the slight difference in the share of population of both communities, Table 7.3

shows that rural households still comprised the highest percent contribution at about 71.10% in

2004 and 74.79% in 2011 in total MEPI. While a 3.69% reduction was posted in the contribution

of the MEPI-poor in urban areas, from 2004 to 2011. This can be attributed to the significant

decline in both MEPI headcount and intensity. As shown in Figure 2.4, convenience and price

were top considerations in a household’s preference of a household of energy source to use. Thus,

rural households stand on a disadvantage due to their financial incapability and geographic

location.

Table 7.3 Energy poverty profile by urbanity, 2004 and 2011


Average
Multidimensional Intensity of Multidimensional
Population Energy Poverty Multidimen- Energy Poverty
Urbanity Headcount (H) sional Energy Index (MEPI)
Poverty (A)
Number Contrib. Ratio Contrib. Ratio Ratio Contrib.
(in
millions)
Urban
2004 8.45 49.81 43.81 35.69 0.51 22.34 28.90
2011 9.15 43.65 35.99 28.48 0.52 18.89 25.21
Rural
2004 8.52 50.19 78.34 64.31 0.70 54.53 71.10
2011 11.82 56.35 70.02 71.52 0.62 43.41 74.79
Source of basic data: HECS 2004
Note: All figures in % except for population

Figure 7.3 accounts for the relative contribution of each energy dimension by urbanity. It

illustrates that the percent contribution of cooking is shown to be significant in 2004 and 2011.

Albeit, a decrease in MEPI headcount is shown for both communities in Table 7.3, there is a posted

significant increase in the use of traditional cooking fuels as shown in Figure 7.3.

44
Figure 7.3. Contribution of MEPI dimensions by urbanity, 2004 and 2011

80.00
Headcount (%) Urban Rural
60.00

40.00

20.00

0.00
2004 2011 2004 2011
Cooking Lighting Services Entertainment and Education

Source of basic data: HECS 2004 & 2011


Note: Values are in terms of percent (%). See Appendix (Table A.1) for values.

Regionally, Table 7.4 and Table 7.5 show a consistent trend of MEPI poor that is

concentrated mostly in ARMM, Zamboanga Peninsula, SOCCSKSARGEN and MIMAROPA.

ARMM posts the highest MEPI at 70.8% in 2004 and 62.84% in 2011 which World Bank (2013)

suggests to be caused by the political instability in the region particularly, the “the struggle of the

Moro groups for self-determination, communist insurgencies, and banditry.” Albeit this, ARMM

only contributes 5.6% to energy poverty in the Philippines vis-a-vis Western Visayas which has

the highest contribution at 51.24%.

Figure 7.4 shows the geographic mapping of MEPI in the Philippines during 2004 and

2011. The shades indicate the concentration of energy poverty in the country—the darker the shade

is, the higher the value of MEPI for that region. The figure illustrates the change of energy poverty

in the Philippines especially in the region of Mindanao and Visayas. In contrast, NCR, Central

Luzon, and CALABARZON consistently ranked the lowest MEPI results in both years due to their

geographical advantage—investments are easily deployed in these regions unlike in Mindanao.

Moreover, the cooking dimension has a consistent trend of having the highest percentage share

through all the regions (see Figure 7.5).

45
Table 7.4 Energy poverty profile by region, 2004
Average
Multidimensional Intensity Multidimensional
Population Energy Poverty of Multidi- Energy Poverty
Region Headcount (H) mensional Index (MEPI)
Energy
Poverty
(A)
Number Contrib. Ratio Contrib. Ratio Ratio Contrib.
(in
millions)
NCR 2.36 13.9 28.19 6.41 0.39 10.90 3.94
CAR 0.29 1.68 71.74 1.97 0.60 43.22 1.89
Ilocos Region 0.89 5.27 59.29 5.11 0.58 34.57 4.73
Cagayan Valley 0.60 3.54 68.80 3.98 0.63 43.18 3.97
Central Luzon 1.87 11.02 52.06 9.38 0.48 24.99 7.15
CALABARZON 2.30 13.54 44.66 9.89 0.52 23.30 8.20
MIMAROPA 0.52 3.04 79.12 3.93 0.72 57.19 4.52
Bicol Region 0.97 5.69 74.76 6.96 0.68 50.50 7.47
Western Visayas 1.29 7.59 77.03 9.56 0.67 51.24 10.10
Central Visayas 1.25 7.36 72.20 8.69 0.67 48.12 9.20
Eastern Visayas 0.77 4.53 75.25 5.58 0.72 54.31 6.39
Zamboanga
0.60 3.55 79.65 4.62 0.75 60.09 5.54
Peninsula
Northern Mindanao 0.76 4.47 69.69 5.10 0.67 46.93 5.45
Davao Region 0.83 4.91 67.68 5.44 0.70 47.25 6.03
SOCCSKSARGEN 0.73 4.33 78.83 5.58 0.73 57.28 6.44
ARMM 0.52 3.08 93.86 4.73 0.75 70.80 5.67
Caraga Region 0.42 2.5 74.97 3.07 0.68 51.09 3.32

Philippines 16.97 100 0.611 100.00 0.63 0.385 100.00


Source of basic data: HECS 2004
Note: All figures in % except for population

46
Table 7.5 Energy poverty profile by region, 2011
Average
Multidimensional Intensity Multidimensional
Population Energy Poverty of Multidi- Energy Poverty
Region Headcount (H) mensional Index (MEPI)
Energy
Poverty
(A)
Number Contrib. Ratio Contrib. Ratio Ratio Contrib.
(in
millions)
NCR 2.72 12.96 18.34 4.31 0.43 7.91 3.13
CAR 0.38 1.79 57.37 1.86 0.59 34.12 1.87
Ilocos Region 1.17 5.59 49.39 5.00 0.53 25.95 4.43
Cagayan Valley 0.78 3.71 61.22 4.12 0.54 33.17 3.76
Central Luzon 2.30 10.97 39.71 7.90 0.49 19.64 6.59
CALABARZON 2.76 13.17 37.20 8.88 0.54 19.94 8.03
MIMAROPA 0.69 3.28 79.29 4.72 0.64 50.62 5.08
Bicol Region 1.19 5.67 74.14 7.62 0.63 46.34 8.03
Western Visayas 1.68 8.01 72.27 10.49 0.58 41.74 10.22
Central Visayas 1.56 7.43 65.42 8.82 0.64 41.57 9.45
Eastern Visayas 0.97 4.61 71.10 5.94 0.62 43.84 6.18
Zamboanga
0.76 3.61 77.50 5.07 0.67 51.90 5.73
Peninsula
Northern Mindanao 0.94 4.50 69.25 5.65 0.62 42.75 5.89
Davao Region 1.01 4.80 66.00 5.75 0.63 41.44 6.09
SOCCSKSARGEN 0.93 4.42 76.04 6.09 0.66 50.46 6.82
ARMM 0.61 2.89 87.78 4.59 0.72 62.84 5.54
Caraga Region 0.54 2.58 68.19 3.19 0.59 40.03 3.16

Philippines 20.97 100.00 0.55 100.00 0.59 0.33 100.00


Source of basic data: HECS 2011
Notes: All figures in % except for population.

47
Figure 7.4. Geographic Mapping of MEPI in the Philippines, 2004 & 2011

48
Figure 7.5. Contribution of MEPI dimensions by region, 2004 and 2011
2004 2011

Caraga
Soccsksargen
Davao Region
ARMM
Mimaropa
Northern Mindanao
Western Visayas
Zamboanga Peninsula
Central Visayas
Bicol Region
Eastern Visayas
Ilocos Region
CAR
Cagayan Valley
Calabarzon
Central Luzon
NCR
45 40 30 25 25 30 35
20 15 10 5 0 0 0 5 10 15 20

Cooking Lighting Services Entertainment and Education


Source of basic data: HECS 2004 & 2011
Note: Values are in terms of percent (%). See Appendix (Table A.2) for values.

The gender and industry sector were also considered despite the limitation in sources of

data. The results show that male-headed households have a higher MEPI value compared to

households that are female-headed. For the industries, fishery and agriculture sectors have the

highest value. However, as this is not the focus of our study, we will not be dwelling much about

this sector. The results that we have found are in the appendix (Table A.3). The contribution of the

MEPI dimensions to gender and the industry sector, are also in the appendix (Tables A.4 and A.5)

where it shows that consistently, the use of traditional fuels for cooking hold majority of the share

for both sectors.

Correlation Tests and Regression Results

To test for the validity of our results, we conducted an OLS regression and a multinomial

logistic regression to further evaluate the values of our index where a relative use of MEPI 2004

with MPI, poverty incidence 2003 was used as data. This is also the case for MEPI 2011 and MPI,

49
poverty incidence 2012. The OLS regression results show that a higher MEPI value (i.e. more

energy poor) leads to a higher poverty incidence level, same with MPI. On the other hand, the

multinomial logistic regression shows that as the income category of households increase, the

probability of having a high MEPI value decreases relative to the highest income category. The

regression results are found in the appendix (Section B.2).

Aside from conducting regression, we also performed relative correlation tests with MPI

and poverty incidence for further evaluation of our MEPI values. The Pearson coefficient of

correlation, Spearman’s rank correlation coefficient, and the Kendall rank correlation coefficient

were used based on the satisfaction of the assumptions on the variables of interest. The results vary

between +1 and -1 in which a variable with a value close or equal to 1 is associated with a high

degree of correlation. Table 7.6 shows a summary of the results of the relative correlation tests that

we employed. The values generated from Spearman’s and Kendall are significant at all

conventional levels, while the result from Pearson correlation emanates a 5% significance level.

They posit a positive correlation with MEPI, which implies that a higher level of MEPI is

associated with higher levels of MPI and poverty incidence. In reference to Cohen’s (1988)

guidelines, the Pearson coefficients emanate a medium strength correlation at 0.404 for MPI and

0.417 for poverty incidence in 2011. Given these results, we then compare MEPI, MPI, and poverty

incidence more specifically.

Table 7.6 Tests for correlation results of MEPI


Spearman’s Rank
Pearson Correlation Kendall Rank Correlation
Correlation
Variable
2004 2011 2004 2011 2004 2011
Name
MPI 0.436* 0.404* 0.433 0.419 0.285 0.272
Poverty
0.417* 0.408* 0.405 0.405 0.266 0.263
Incidence
Note: *values are in terms of 0.05 significance level; all the rest are significant at all levels.

50
Comparison of MEPI with other poverty measures

Figure 7.6 compares and shows the value of poverty in terms of MEPI (as computed

above), multidimensional poverty index (MPI), income headcount and poverty incidence for the

years 2003 to 2004 and 2011 to 2012. The values of MPI and income headcount are computed by

using FIES dataset for 2003 and 2012, following Balisacan’s weights allocation and methodology.

The graph shows that consistently, ARMM shows the highest ratio for MEPI and MPI and income

headcount in 2011-2012 at an average of 67%. This was consistently followed by Zamboanga

Peninsula at 51.90% and 23.54% for MEPI and MPI of 2011-2012. But for the poverty incidence,

ARMM was followed by Eastern Visayas where 45.2% of the population are below the poverty

threshold of the region during 2012. Zamboanga Peninsula only came in as the region with the 6th

highest number of percentage in terms of the poverty incidence.

One can also see that MEPI and MPI moves in the same direction, on the other hand

however, income headcount and poverty incidence sometimes show a different movement for both

years. This further shows the high discrepancy between the numbers generated through the

unidimensional approach (poverty threshold) in comparison to the multidimensional approach

(MPI).

For further analysis, the MPI of Balisacan (2015) was deconstructed by tweaking the

standard of living dimension to account for the contribution of energy to the overall MPI (see Table

7.7). The “Energy” dimension was created accounts for electricity access, utilization of modern

energy sources and ownership of recreational appliances. The reconstructed table with modified

indicators and weights is found in the appendix Table A.8. Moreover, the Family Income and

Expenditures Survey (FIES) for 2003 and 2012 were utilized for the said computations to allow

relative comparison for MEPI 2004 and 2011 computed using HECS. The key results in running

the reconstructed MPI report that the energy dimension shows a minimal contribution to the overall
51
poverty at about 32.76% for 2003 and 37.86% for 2012, in the Philippines. Surprisingly, energy

dimension’s contribution to MPI is highest in Ilocos Region followed by Cagayan Valley and

CALABARZON if FIES is used. This is relatively different with the findings of MEPI using HECS

2011 where areas in Mindanao comprise the most contribution to the MEPI of the Philippines. It

is important to note that MEPI contribution is computed by looking at which region/urbanity

contributes the most to the MEPI of the Philippines compared to the energy contribution column

where it looks at the influence of the energy dimension to the overall MPI in terms of region.

Figure 7.6. Comparison of MEPI, MPI, income headcount and poverty incidence by region,
2003-2004 and 2011-2012
2003 - 2004
MEPI MPI Income Headcount Poverty Incidence
80.00
70.00
PERCENTAGE

60.00
50.00
40.00
30.00
20.00
10.00
0.00

2011 -2012
MEPI MPI Income Headcount Poverty Incidence
70.00
60.00
PERCENTAGE

50.00
40.00
30.00
20.00
10.00
0.00

Source of basic data: HECS 2004 & 2011 for MEPI, FIES 2003 & 2012 for MPI and Income Headcount,
Philippine Statistics Yearbook 2016 for Poverty Incidence
Note: 2003 &2012 MPI & Income Headcount, and 2004 & 2011 MEPI are based on author’s
computation. See Tables A.6 and A.7 for values.
52
Table 7.7 Contribution of energy dimension to MPI, 2003 and 2012
Subgroup Health Education Standard of living Energy MEPI
Urbanity 2003 2012 2003 2012 2003 2012 2003 2012 2004 2011
Urban 17.84 21.44 8.74 13.84 40.26 24.66 33.16 40.05 28.90 25.21
Rural 21.38 26.89 10.23 12.50 34.00 23.49 34.39 37.13 71.10 74.79

Region
NCR 12.41 16.49 20.35 17.56 39.68 25.44 27.56 40.51 3.94 3.13
CAR 20.94 27.23 12.98 13.33 32.06 21.02 34.02 38.42 1.89 1.87
Ilocos Region 21.44 22.78 9.49 7.49 36.14 25.42 32.93 44.31 4.73 4.43
Cagayan Valley 18.34 20.93 11.61 11.91 35.58 24.46 34.47 42.70 3.97 3.76
Central Luzon 19.84 21.00 13.88 14.44 36.82 22.87 29.46 41.69 7.15 6.59
Calabarzon 20.01 20.53 12.55 12.88 36.75 24.81 30.69 41.78 8.20 8.03
Mimaropa 20.04 26.80 11.33 13.96 34.51 24.00 34.12 35.24 4.52 5.08
Bicol Region 23.08 28.34 9.73 11.26 34.05 24.03 33.14 36.36 7.47 8.03
Western Visayas 21.52 24.81 11.16 10.57 33.75 25.82 33.57 38.80 10.10 10.22
Central Visayas 24.08 25.74 12.03 11.85 31.83 24.14 32.05 38.26 9.20 9.45
Eastern Visayas 20.88 29.05 13.18 14.08 33.18 23.87 32.76 32.99 6.39 6.18
Zamboanga Peninsula 23.37 25.16 12.04 13.75 32.20 23.94 32.39 37.15 5.54 5.73
Northern Mindanao 23.35 27.85 12.77 12.72 30.41 21.70 33.48 37.73 5.45 5.89
Davao Region 20.86 25.20 14.43 14.16 31.73 23.33 32.98 37.31 6.03 6.09
Soccsksargen 20.94 26.96 11.58 13.43 33.40 23.49 34.09 36.12 6.44 6.82
ARMM 22.32 31.01 9.68 13.99 33.49 21.37 34.52 33.63 5.66 5.54
Caraga Region 21.46 25.27 12.37 14.12 32.89 22.95 33.27 37.67 3.32 3.16
Philippines 21.46 25.52 12.05 12.84 33.73 23.78 32.76 37.86 100.00 100.00
Source of basic data: FIES 2003 and FIES 2012 for MPI; HECS 2004 and 2011 for MEPI
Notes: MPI based on author’s computation and energy indicators were removed in standard of living as MPI derived from Balisacan (2015). See
Appendix (Table A.8) for the weights and indicators used by the authors for this computation.

53
Table 7.8 shows the trend of varying energy poverty cut-off (k) values to MEPI, average

energy poverty intensity (A) and energy poverty headcount (H). As seen in the tabular summary

below, as the k value increases, MEPI and H display a decreasing trend while A exhibits a

proportional increasing trend. This verifies that as more dimensions are taken into account, the

depth (A) of energy poverty intensifies and the number people deprived (H) drop. The results

indicate that the overall energy headcount has an average decrease of 8% relative to the reduction

in average energy intensity of 2.1%. Overall, MEPI decreased by an average of 6.2%.

Table 7.8 MEPI under different k values


Multidimensional Energy Poverty Indexes
MEPI A H
k 2004 2011 2004 k 2004 2011
0.1 0.428 0.365 0.508 0.1 0.428 0.365
0.2 0.418 0.348 0.542 0.2 0.418 0.348
0.3 0.385 0.327 0.630 0.3 0.385 0.327
0.4 0.339 0.281 0.717 0.4 0.339 0.281
0.5 0.308 0.234 0.762 0.5 0.308 0.234
0.6 0.274 0.203 0.804 0.6 0.274 0.203
0.7 0.182 0.143 0.906 0.7 0.182 0.143

Figure 7.7 illustrates the test of first-order dominance for the MEPI. This tests for the

robustness of the change in poverty in a per period basis. Non-intersecting lines represent

robustness of results in a particular poverty cut-off. The graph shows plausible results as they pass

the robustness tests at all poverty cut-offs.

54
Figure 7.7. Trend of MEPI in different poverty cut-offs

45%

MEPI ( Adjusted Headcount)


40%
35%
30%
25%
20%
15%
10%
5%
0%
0 0.2 0.4 0.6 0.8
Poverty cutoff k

HECS 2004 HECS 2011

55
CHAPTER 8: Conclusion and Policy Implications

This study shows the extent of energy poverty in the Philippines through the

Multidimensional Energy Poverty Index (MEPI). We found that energy poverty in a

multidimensional lens exhibit a decreasing trend as fuelled by significant reduction in the depth

and breadth of energy poverty. Furthermore, the use of fuelwood as a major source for cooking

supported the findings of this research as for the dominance in the use of traditional fuels for

cooking which further aggravates index. Furthermore, effects of the rural energy programs started

in 1995 were not yet felt as energy poverty has still been reported as a rural phenomenon and the

contribution of having no electricity access surged above 1% in 2011. MEPI-poor are concentrated

in the areas of Mindanao specifically, ARMM, Zamboanga Peninsula, SOCCSKSARGEN and

MIMAROPA, while NCR, followed closely by Central Luzon, yields the lowest percent

contribution to overall MEPI. Male-headed households are also deemed to be more MEPI-poor

than the female-headed households and that a portion of energy poverty came largely from the

agriculture and fishery sector which involves low-skilled and low-paying occupations. Thus, this

ponders the need for further electrification programs and off-grid expansions to rural villages.

Herrin (1979) discovered that rural electrification along with other social and economic

developments bring about changes in consumption and investment patterns of rural households.

Additionally, Chakravorty et. al. (2016) shows that rural electrification raises agricultural income

and has significant short term benefits for regions in low income nations such as the Philippines.

Given the following results obtained through this research, this study may serve as an

impetus to strengthen efforts in providing modern energy services such as electricity, LPG and

kerosene, to regions that highly exhibits energy poverty as obtained through MEPI such as in

ARMM, Zamboanga Peninsula, and SOCCSKSARGEN. Partnerships and regular dialogues to

56
key stakeholders such as the electric power industry participants, and LGUs can also be established

to facilitate policy recommendations and program developments. Continuous work on policies can

also be done to attract private capital in power generation, and ease up the procedures in obtaining

permits and licenses among investors. Furthermore, stronger efforts in conducting studies that

would further provide knowledge about the importance of energy access to welfare development

and poverty reduction such as the Household Energy Consumption Survey (HECS) can also be

conducted, and use this as a foundation in laying out effective energy policies and programs.

We also encourage further research about the state of energy poverty in the Philippines

using a different approach of measuring energy poverty and/or through using a different source of

data to be able to analyse the status of energy poverty in a different perspective. We recommend

future researchers to also look upon the different determinants of MEPI and how it may affect the

macroeconomy (i.e. economic growth, poverty) by employing different economic regression

models using MEPI as an independent variable.

57
BIBLIOGRAPHY

Asian Development Bank (ADB). 2005. “Assessing the Impact of Transport and Energy
Infrastructure on Poverty Reduction.” Manila.

Albis, M., and Elviña, J. 2017. “Employment Correlates of Multidimensional Poverty in the
Philippines.” Working Paper No. 2017-01, UP School of Statistics

Alkire, S. 2002. “Dimensions of Human Development.” World Development, 30(2), 181–205.

Alkire, S. and Santos, M. E. 2010. “Acute multidimensional poverty: A new index for development
countries.” Oxford Poverty and Human Development Initiative (OPHI) Working Paper 38,
Oxford Department of International Development, University of Oxford.

Alkire, S. and Foster, J. 2007. “Counting and multidimensional poverty measures.” Oxford Poverty
and Human Development Initiative (OPHI), Working Paper 7, Oxford Department of
International Development, University of Oxford.

___ 2009. “Counting and Multidimensional Poverty Measurement.” OPHI Working Paper, No.
32, University of Oxford.

___2011. “Counting and Multidimensional Poverty Measurement.” Journal of Public Economics,


95, 476–487.

Anand, S. and Sen, A. 1997. “Concepts of human development and poverty: a multidimensional
perspective.” Human Development Papers.

Antony, G. M., and Rao, K.V. 2007. “A composite index to explain variations in poverty, health,
nutritional status and standard of living: use of multivariate statistical methods.” Public
Health, 121, 578-587.

Balisacan, A. M. 2001. “Pathways of Poverty Reduction: Rural Development and Transmissions


Mechanisms in the Philippines.” Paper presented at the Asia and Pacific Forum on Poverty:
Reforming Policies and Institutions for Policy Reductions. Manila, 5-9 February.

Balisacan, A. M. 2015. “The Growth-Poverty Nexus: Multidimensional Poverty in the


Philippines.” In Balisacan, A., Chakravorty, U., & Ravago, M. L. (Eds.). Sustainable
Economic Development: Resources, Environment, and Institutions (pp. 445-465). Elsevier.

58
Barnes, D.F., Khandker, S., Samad, H. 2010. “Energy poverty in rural Bangladesh” on Energy
Policy, 39, 894-904. doi: 10.1016/j.enpol.2010.11.014

Barnes, D. F. 2010. “The Concept of Energy Poverty, Energy for Development and Poverty
Reduction”. http://www.energyfordevelopment.com/2010/06/energy-poverty.html

Bersisa, M. 2016. “Multidimensional Measure of Household Energy Poverty and its Determinants
in Ethiopia”. East Africa Research Papers in Economics and Finance.

Binswanger, H. and Khandkher S. 1993. “How infrastructure and financial institutions affect
agricultural output and investment in India” In Journal of Development Economics, 41, (2),
337-366. doi: 10.1016/0304-3878(93)90062-R

Bourguignon, F., and Chakravarty, S. 2003. “The Measurement of Multidimensional poverty.”


Journal of Economic Inequality, 1, 25–49.

Bravo, V., Mendoza, G., Legisa, J., Suarez, C.E., and Zyngierman, I. 1979. Estudio Sobre
Requerimientos Futuros No Convencionales de Energia en America Latina, Project
RLA/74/030, Fundacion Bariloche, Buenos Aires, Argentina, Report to the United Nations
Development Program, 1979, Appendix 9, Primera Aproximación a una Definición de las
Necesidades Basicas.

Calvo, C. 2008. “Vulnerability to multidimensional poverty: Peru, 1998-2002.” World


Development, 36(6), 1011-20.

Casimiro, G. and Ballester, R.E. 2013. “A Multidimensional Approach to Chid Poverty in the
Philippines.” Paper presented at the 12th National Convention on Statistics, EDSA Shangrila
hotel, Mandaluyong.

Chakravorty, U. Emerick, K. and Ravago, M-L. 2016. “Lighting Up the Last Mile: The Benefits
and Costs of Extending Electricity to the Rural Poor”. Working Paper 2016-04, Energy
Policy and Development Program (EPDP)

Chiappero-Martinetti, E. 2000. “A multidimensional assessment of well-being based on Sen's


functioning approach.” Rivista Internazionale di Scienze Sociali, 108(2), 207-239.

Clancy, J. 2003. “Household energy and gender: the global context”.

59
Cohen, J. 1988. Statistical Power Analysis for the Behavioral Sciences, 2nd ed. Hillsdale, NJ:
Erlbaum.
Datt, G. 2016. “Multidimensional Poverty in the Philippines, 2004-13: Do choices for weighting,
identification and aggregation matter?”

Department of Energy (DOE). 2012. “Philippine Energy Plan 2012-2030”. Philippines.

Energy Sector Management Assistance Program (ESMAP). 2002. “Rural Electrification and
Development in the Philippines: Measuring the Social and Economic Benefits”. Washington
D.C.

Ediger, V. and Tathdil, H. “Energy as an Indicator of Human Development: A Statistical


Approach” In Journal of Energy and Development, 31, 213-232.

Edoumiekumo, S.G., Tombofa, S., and Karimo, T.M. 2013. “Multidimensional Energy Poverty in
the South-South Geopolitical Zone of Nigeria”. Journal of Economics and Sustainable
Development, 4(20), 96-103.
Fan, S., Zhang, L., and Zhang, X. 2002. “Growth, Inequality, and Poverty in Rural China: The
Role of Public Investment”. Research Report 125. International Food Policy Research
Institute. Washington D.C.
Fan, S., Jitsuchon, S., and Methakunnavut, N. 2004. “The Importance of Public Investment for
Reducing Rural Poverty in Middle-Income Countries: The Case of Thailand”. International
Food Policy Research Institute DSGD Discussion Paper 7.
Fan, S., Nyange, D., and Rao, N. 2005. “Public Investment and Poverty Reduction in Tanzania:
Evidence from Household Survey Data”. International Food Policy Research Institute DSGD
Discussion Paper 18.
Foster, V., Tre, J-P. and Wodon, Q. 2000. “Energy Prices, Energy Efficiency, and Fuel Poverty”.
Washington, DC: Latin America and Caribbean Regional Studies Programme, the World
Bank.

Gordon, D., Levitas, R., Pantazis, C., Patsios, D., Payne, S., and Townsend, P. 2000. “Poverty and
social exclusion in Britain”. Joseph Rowntree Foundation, York

Gordon, D., Nandy, S., Pantazis, C., Pemberton, S., and Townsend, P. 2003. “The Distribution of
Child Poverty in the Developing World.” Bristol: Centre for International Poverty Research.

60
Grogan, L. 2008. “Community Electrification and Labour Market Development”. Working paper,
Department of Economics, University of Guelph.

Haughton J. and Khandker, S.R. 2009. “Handbook on Poverty and Inequality.” Washington, DC:
World Bank.

Herrin, A. “Rural Electrification and Fertility Change in the Southern Philippines” in Population
and Development Review, 5 (1): 61-86

International Energy Agency (IEA). 2004. World Energy Outlook 2004. Paris: Organisation for
Economic Co-operation and Development (OECD)/IEA.

___ 2010. World Energy Outlook 2010. Paris: OECD/IEA.

___ 2013. Southeast Asia Energy Outlook 2013. Paris: OECD/IEA.

___ 2014a. Energy Poverty. Paris: OECD/IEA. http://www.iea.org/topics/energypoverty/

___ 2015. Southeast Asia Energy Outlook 2015. Paris: OECD/IEA.

Lipton, M., and Ravallion, M. 1995. “Poverty and policy” on Handbook of development
economics, 3, 2551-2657

Kanagawa, M. and T. Nakata. 2008. “Assessment of access to electricity and the socio- economic
impacts in rural areas of developing countries.” Energy Policy, 36, 2016-2029. doi:
10.1016/j.enpol.2008.01.041

Khandker, S., Barnes, D., and Samad, H. 2009. “Welfare impacts of rural electrification a case
study from Bangladesh.” Washington, DC, World Bank.

___ 2012. “Are the energy poor also income poor? Evidence from India” In Energy Policy, 47, 1-
12.

Kooijman-van Dijk, A. 2008. “The Power to Produce - The Role of Energy in Poverty Reduction
Through Small Scale Enterprises in the Indian Himalayas.” University of Twente. Enschede.

Li, K., Lloyd, B., Liang, X-J., and Wei, Y-M. 2014. “Energy Poor or Fuel Poor: What are the
differences?” In Energy Policy, 68, 476-481. doi: 10.1016/j.enpol.2013.11.012

61
Martınez, D. and Ebenhack, B. 2008. “Understanding the role of energy consumption in human
development through the use of saturation phenomena” In Energy Policy, 36, 1430–1435.
Modi, V., McDade, S., Lallement, D., and Saghir, J. 2005. “Energy Services for the Millennium
Development Goals”. New York: Energy Sector Management Assistance Programme,
United Nations Development Programme, UN Millennium Project, and World Bank.

Mohanty, S. K. 2011. “Multidimensional poverty and child survival in India.” PLoS ONE, 6(10),
doi. e26857.

Morales, L. T. 2013. “Philippines Country Report” in Kimura, S. (ed.), Analysis on Energy Saving
Potential in East Asia. ERIA Research Project Report 2012-19, 255-275.

Navarro, A. 2012. “Finding Solutions to the Mindanao Electric Power Problem”. Discussion Paper
No. 2012-21, PIDS

Nussbaumer, P., Bazilian, M., Modi, V., and Yumkella, K. 2011. “Measuring Energy Poverty:
Focusing on What Matters”. Renewable and Sustainable Energy Reviews, 16 (1), 231-243.

Nussbaumer, P., Nerini, F.F., Onyeji, I., and Howells, M. 2013. “Global Insights Based on the
Multidimensional Energy Poverty Index (MEPI)”. Sustainability, 2013 (5), 2060-2076.

OECD and IEA. 2010. “World Energy Outlook”. Paris.

Pachauri, S. and Spreng, D. 2004. “Energy Use and Energy Access in Relation to Poverty”.
Economic and Political Weekly, 39(3), 271-278.

Parajuli, R. 2011. “Access to energy in Mid/Far west region-Nepal from the perspective of energy
poverty”. Renewable Energy, 36, 2299-2304.

Pasternak, A. 2000. “Global Energy Futures and Human Development: A Framework for
Analysis”. US Department of Energy Report UCRL-ID- 140773, Lawrence Livermore
National Laboratory, Livermore, CA.
Pereira, M.G., Freitas, M.A.V., and da Silva, N.F. 2011. “The challenge of energy poverty:
Brazilian case study” In Energy Policy, 39, 167–175.

Philippine Statistics Authority (PSA). 2009—2016. “Philippine Statistics Yearbook”.

62
Pokharel S. 2007. “Kyoto protocol and Nepal's energy sector.” In Energy Policy 35(4): 2514-2525.
doi: 10.1016/j.enpol.2006.09.015

Power for All. 2014. “The Energy Access Imperative”.

Qi, D. and Wu, Y. 2015. “A multidimensional child poverty index in China”. On Child Youth Serv
Rev 57:159–170. doi:10.1007/s12187-015-9351-1

Qizilbash, M. 2004. “On the arbitrariness and robustness of multi-dimensional poverty rankings.”
UNU-WIDER Research Paper (2004/37).

Reiche, K., Covarrubias, A., and Martinot, E. 2000. “Expanding electricity access to remote areas:
off-grid rural electrification in developing countries.” Fuel, 1, 1.4. doi: 10.1186/s13705-
015-0037-9

REN21. 2013. Renewables 2013 Global Status Report.

Schaffner, J. 2014. Development Economics.

Sen, A. 1985. “Commodities and Capabilities.” North-Holland.

Sher, F., Abbas, A., Awan, R.U. 2014. “An Investigation of Multidimensional Energy Poverty in
Pakistan: A Province Level Analysis”. In International Journal of Energy Economics and
Policy, 4(1), 65-75.
Sihag AR., Misra, N., and Sharma, V. 2004. “Impact of power sector reform on the poor: case-
studies of South and South-East Asia.” Energy. Sust. Dev., 8(4): 54-73. doi: 10.1016/S0973-
0826(08)60513-X

Tchereni, B.H.M., Grobler, W., and Dunga, S. 2013. “Economic Analysis of Energy Poverty in
South Lunzu, Malawi”. In Journal of Economics and Sustainable Development, 4(4), 154-
163.

United Nations (UN), 2014. “Report of the Open Working Group of the General Assembly on
Sustainable Development Goals (Report No. A/68/970).” Retrieved from:
http://undocs.org/A/68/970

63
___ 2015. “The Millenium Development Goals Report 2015.” Retrieved from:
http://www.un.org/millenniumgoals/2015_MDG_Report/pdf/MDG%202015%20re
v%20(July%201).pdf

UNDP 2010. “Human Development Report 2010.” New York: Palgrave Macmillan.

___ 2013. “Gender Energy” Gender and Climate Change Capacity Development Series Asia and
the Pacific.

Wagle, U. R. 2008. “Multidimensional poverty: an alternative measurement approach for the


United States? Social Science Research.” 37, 559-580.

Wang, X. L. and Alkire, S. 2009. “Measurement of Multidimensional Poverty in China: Estimation


and Policy Implications.” Chinese Rural Economy, No.12, 2009, 4–23. Sabina Alkire, 2009,
4–23 ).

Wang, Y. and Wang, B. 2016. “Multidimensional Poverty Measure and Analysis: Case Study from
Heichi City, China”

World Bank Group. 2014. “Capturing the Multi-Dimensionality of Energy Access”

World Bank. 2000. “World Development Report 2000/2001: Attacking Poverty.” Washington, DC:

___ 2008. “The Welfare Impact of Rural Electrification: A Reassessment of the Costs and
Benefits.” An IEG Impact Evaluation, Washington, D.C.

64
APPENDIX

A. Tabular Values of Data and Results


Figure A.1. Percent of households using traditional fuel by end-use

Fuelwood Charcoal Biomass Residues

99.8
99.6

90.1
88.1

97
85.9
67.2

63.4

30.7
27.6
37
25.9
23.4
16.6

11.2

10.5
8.7

6.7
4.9
3.8

3.5
1989 1995 2004 2011 1995 2004 2011
C OOKING WATE R HE ATING

Source of basic data: 1989 Energy Statistics, 1995, 2004, and 2011 HECS by NSO-DOE
Notes: Data for water heating for 1989 is unavailable. A household may be counted for more than one
under each type of fuel and end-use.

Figure A.2. Households deprived by end-use and electricity access (in %)


Refrigeration Space Cooling Entertainment and Education Electricity Access
67.4

58.62
55.1
49.6
35.3

27.61
25.4
19.4
16.1

11.81
12.8
12.4
10.2
0
0
0

1989 1995 2004 2011


DE PR IVE D

Source of basic data: 1989 Energy Statistics, 1995, 2004, and 2011 HECS by NSO-DOE
Notes: Data for refrigeration, space cooling and entertainment and education for 1989 is unavailable.
Entertainment and education deprivation is indicated as a proxy for recreational activities deprivation.
*A household may be counted for more than one under each type of fuel and end-use.

65
Figure A.3. Households deprived of MEPI indicators by industry sector, 2011

100
Percentage 80
60
40
20
0

Electricity Access Modern Cooking Sources


Services Space Cooling
Entertainment and Education

Source of basic data: HECS 2004 and 2011

Figure A.4. Households deprived of MEPI indicators by gender, 2011

80

60
Percentage

40

20

0
Male Female
Electricity Access Modern Cooking Sources Services
Space Cooling Entertainment and Education

Source of basic data: HECS 2004 and 2011


Table A.1 Energy poverty contribution per dimensions by urbanity

Service provided
by means of Entertainment and
Community Cooking Lighting
Household Education
Type
Appliances
Urban
2004 9.72 2.02 14.72 9.41
2011 14.69 2.43 11.73 8.31

Rural
2004 21.86 7.17 14.13 29.76
2011 24.33 6.60 12.49 29.47
Note: All values are in percent (%).

66
Table A.2 Energy poverty contribution per dimensions by region (2004 & 2011)

Regions Cooking Lighting Services Entertainment


Provided by and Education
Means of
Household
Appliances

2004 2011 2004 2011 2004 2011 2004 2011


NCR 6.06 8.70 0.18 1.79 3.91 2.36 0.75 2.89
CAR 20.26 16.47 2.49 2.97 16.56 14.99 3.91 3.69
Ilocos Region 18.01 18.57 2.34 1.75 11.68 12.66 2.53 2.29
Cagayan Valley 20.45 18.07 3.69 2.08 14.90 12.84 4.13 2.58
Central Luzon 13.40 13.54 1.21 1.09 8.83 8.85 1.55 1.56
Calabarzon 12.58 11.07 0.95 4.10 8.03 5.54 1.73 6.74
MIMAROPA 27.95 21.11 4.16 8.35 18.31 10.25 6.77 14.51
Bicol Region 27.02 24.28 2.81 3.82 16.41 15.89 4.26 4.33
Western Visayas 28.84 24.02 2.92 3.08 16.38 16.66 3.10 3.85
Central Visayas 25.58 22.89 3.08 3.97 16.11 15.71 3.35 4.23
Eastern Visayas 26.92 22.81 4.19 3.01 17.75 15.33 5.45 4.39
Zamboanga Peninsula 30.00 24.41 5.32 6.01 19.33 19.33 5.44 6.15
Northern Mindanao 23.40 23.81 3.51 4.13 16.02 18.28 4.00 4.41
Davao Region 24.57 23.07 2.67 5.04 15.20 18.02 4.81 5.34
SOCCSKSARGEN 29.59 26.97 4.09 4.89 17.79 17.40 5.82 4.84
ARMM 31.35 25.10 6.88 7.83 23.89 21.06 8.67 7.21
Caraga Region 25.34 22.78 4.00 3.93 17.35 20.10 4.39 5.04
Note: All values are in percent (%).

67
Table A.3 Energy poverty profile by gender & industry sector, 2011
Average
Multidimensional Intensity of Multidimensional
Population Energy Poverty Multidimen- Energy Poverty
Sector Headcount (H) sional Energy Index (MEPI)
Poverty (A)
Number Contrib. Ratio Contrib. Ratio Ratio Contrib.
(in
millions)
Gender*
Male 16.69 79.59 56.95 82.16 37.55 33.88 82.45
Female 4.28 20.41 48.21 17.84 32.40 28.12 17.55

Industry*
Agriculture 5.57 26.56 83.22 40.07 56.87 55.00 44.67
Fishery 0.90 4.29 86.94 6.76 59.92 58.06 7.62
Mining 0.12 0.55 74.90 0.75 53.45 49.88 0.84
Manufacturing 1.12 5.34 42.77 4.14 27.23 22.40 3.65
Utilities 0.07 0.35 17.97 0.11 16.71 11.57 0.12
Construction 1.49 7.12 56.36 7.28 34.55 30.08 6.55
Trade 2.09 9.95 37.31 6.73 24.03 18.94 5.76
Transportation
1.90 9.04 43.74 7.17 27.29 22.22 6.14
and ICT
Finance 0.11 0.52 14.21 0.13 11.53 5.73 0.09
Services 3.17 15.12 37.16 10.18 23.73 18.91 8.74
Unemployed 4.43 21.15 43.48 16.67 28.57 24.44 15.80
55.17 100.00 59.29 32.71 100.00
Philippines 20.97 100
Source of basic data: HECS 2011
Notes: All figures in % except for population.
*Gender and industry sector of household head is available only for HECS 2011.

68
Table A.4 Energy poverty contribution per dimensions by business sector, 2011

Services
Provided by
Business Sector Cooking Lighting Means of Entertainment
Household and Education
Appliances
Agriculture 39.24 9.16 29.82 8.73
Fishery 38.49 7.51 28.24 8.98
Mining 31.26 8.81 26.59 8.25
Manufacturing 28.50 2.59 20.71 4.56
Utilities 23.53 3.59 13.54 3.08
Construction 22.58 2.19 15.61 3.10
Trade 23.00 1.63 15.17 2.97
Transportation and ICT 20.00 1.67 13.25 2.39
Finance 25.68 0.00 10.90 0.58
Services 9.49 0.88 6.30 1.31
Unemployed 7.28 0.82 4.92 1.18
Note: All values are in percent (%).

Table A.5 Energy poverty contribution per dimensions by gender, 2011

Services
Provided by
Means of
Household Entertainment
Gender Cooking Lighting Appliances and Education
Male 16.42 2.71 11.77 2.99
Female 13.74 2.01 9.77 2.61
Note: All values are in percent (%).

69
Table A.6 Comparison of MEPI, MPI, income headcount and poverty incidence by region, 2003-2004

HECS 2004 FIES 2003 FIES 2003 PSA 2003


Multidimensional
Multidimensional
Energy Poverty Income Headcount Official Poverty Measure
Poverty Index (MPI)
Index (MEPI)
Annual Per Poverty
Capita Poverty incidence
Contrib. Contrib. Contrib.
Region Ratio Ratio Ratio Threshold (in Among
(%) (%) (%)
Pesos) Population (%)
NCR 10.90 3.94 4.42 3.15 3.32 1.77 16,737 6.90
CAR 43.22 1.89 18.54 1.71 23.03 1.59 14,033 32.20
Ilocos Region 34.57 4.73 12.38 3.44 23.49 4.87 13,281 30.20
Cagayan Valley 43.18 3.97 15.50 2.79 19.71 2.65 11,417 24.50
Central Luzon 24.99 7.15 8.69 5.09 12.53 5.48 14,378 17.50
Calabarzon 23.30 8.20 9.69 6.68 12.88 6.63 14,720 18.40
MIMAROPA 57.19 4.52 28.21 4.54 36.76 4.42 12,402 48.10
Bicol Region 50.50 7.47 27.55 8.85 46.04 11.03 12,379 48.50
Western Visayas 51.24 10.10 25.73 10.37 30.95 9.31 12,291 39.20
Central Visayas 48.12 9.20 25.59 9.98 37.23 10.83 9,805 28.30
Eastern Visayas 54.31 6.39 30.56 7.68 38.50 7.22 10,804 43.00
Zamboanga Peninsula 60.09 5.54 36.00 6.82 46.92 6.63 10,407 49.20
Northern Mindanao 46.93 5.45 24.00 5.71 38.87 6.90 11,605 44.00
Davao Region 47.25 6.03 23.07 5.99 30.75 5.96 11,399 34.70
SOCCSKSARGEN 57.28 6.44 26.66 6.13 34.29 5.89 11,328 38.40
ARMM 70.80 5.66 43.60 7.52 33.41 4.30 12,733 52.80
Caraga Region 51.09 3.32 25.80 3.54 44.20 4.52 11,996 54.00
Philippines 38.49 100.00 18.90 100.00 25.32 100.00 12,309 30.00
Note: FIES 2003 MPI is based on the authors’ computation, and the official poverty measure was derived from the Philippine Statistics Yearbook
(2016).

70
Table A.7 Comparison of MEPI and MPI, income headcount and poverty incidence by region, 2011-2012
HECS 2011 FIES 2012 FIES 2012 PSA 2012
Multidimensional
Multidimensional
Energy Poverty Index Income Headcount Official Poverty Measure
Poverty Index (MPI)
(MEPI)
Annual Per Poverty
Capita incidence
Contrib. Contrib. Contrib. Poverty Among
Region Ratio Ratio Ratio
(%) (%) (%) Threshold Population (%)
(in Pesos)
NCR 7.91 3.13 2.71 2.63 5.33 2.57 20,344 3.90
CAR 34.12 1.87 11.86 1.60 20.52 1.37 19,483 22.80
Ilocos Region 25.95 4.43 6.26 2.45 20.15 3.91 18,373 18.50
Cagayan Valley 33.17 3.76 8.64 2.30 17.62 2.33 19,125 22.10
Central Luzon 19.64 6.59 5.65 4.82 14.55 6.15 20,071 12.90
Calabarzon 19.94 8.03 5.64 6.06 12.43 6.63 19,137 10.90
MIMAROPA 50.62 5.08 21.20 4.93 33.03 3.81 17,292 31.00
Bicol Region 46.34 8.03 19.03 8.69 44.15 10.00 18,257 41.10
Western Visayas 41.74 10.22 16.24 9.59 29.11 8.52 18,029 29.10
Central Visayas 41.57 9.45 15.76 9.00 35.71 10.11 18,767 30.20
Eastern Visayas 43.84 6.18 20.36 6.98 44.73 7.60 18,076 45.20
Zamboanga Peninsula 51.90 5.73 23.54 6.81 37.27 5.35 18,054 40.10
Northern Mindanao 42.75 5.89 16.58 6.08 39.38 7.17 19,335 39.50
Davao Region 41.44 6.09 15.76 5.96 29.87 5.60 19,967 30.70
SOCCSKSARGEN 50.46 6.82 22.52 7.85 43.88 7.59 18,737 44.70
ARMM 62.84 5.54 40.11 10.97 54.32 7.37 20,517 55.80
Caraga Region 40.03 3.16 16.11 3.29 38.91 3.95 19,629 40.30
Philippines 32.71 100.00 12.91 100 26.04 100% 18,935 25.20
Note: FIES 2012 MPI is based on author’s computation, and the official poverty measure was derived from the Philippine Statistics Yearbook
(2016).

71
Figure A.5 Geographic Mapping of MPI in the Philippines, 2003 & 2012

2003 2012

72
Figure A.6 Geographic Mapping of Poverty Incidence in the Philippines, 2003 & 2012

2003 2012

73
Table A.8 Description of modified dimensions, weights, and their cut-off points for
computing MPI
Dimension Indicator Weights Deprivation Cut-off (Poor if…)
Child mortality 1/16 household does not use flush toilet
Sanitation

Drinking 1/16 household does not have access to safe drinking


Health
Water water

Malnutrition 1/8 household is food poor


Food Poverty
Years of 1/41 no household member has completed six years of
Education schooling schooling.

Shelter household’s roof is composed of light/ salvaged


1/24
Roof material

Wall household’s wall is composed of light/ salvaged


1/24 material
Standard of
Mobility
Living household does not own a vehicle
Ownership of
1/12
vehicle
household’s other sources of income are less than
Other sources of 20% of total income
income 1/12
Electricity 1/16 household does not have access to electricity

Modern Energy 1/16 household uses modern energy

HH Appliance 1/32 household has an air conditioning unit


Ownership
Energy (Space Cooling &
Refrigerator 1/32 household has a refrigerating unit
Ownership)

Entertainment 1/16 household has either a personal computer or radio or


and Education both

74
B. Stata Runs
B.1 MEPI Computation
B.1.1 Philippines
2004

Headcount (H)

Average Intensity (A)

75
Multidimensional Energy Poverty Index (MEPI)
. gen mepi=deprive*sumf

2011
. svyset [pw=popwgt]

pweight: popwgt
VCE: linearized
Single unit: missing
Strata 1: <one>
SU 1: <observations>
FPC 1: <zero>

Headcount (H)

Average Intensity (A)

76
MEPI
. gen mepi=deprive*sumf

B.1.2 By Region
svyset [pw=popwgt]
svy: mean deprive, over (reg)
svy: mean sumf if deprive==1, over (reg)
Svy: mean mepi, over (reg)

B.1.3.Industry Sector
svyset [pw=popwgt]
svy: mean deprive, over (dbusiness)
svy: mean sumf if deprive==1, over (dbusiness)
Svy: mean mepi, over (dbusiness)
B.1.4 Gender
svyset [pw=popwgt]
svy: mean deprive, over (headsex)
svy: mean sumf if deprive==1, over (headsex)
Svy: mean mepi, over (headsex)

77
B.2 Regression Results
2004
OLS Regression
. regress pov_inc mepi mpi

Source SS df MS Number of obs = 21960


F( 2, 21957) =53356.79
Model 3724417.65 2 1862208.82 Prob > F = 0.0000
Residual 766322.701 21957 34.9010657 R-squared = 0.8294
Adj R-squared = 0.8293
Total 4490740.35 21959 204.505686 Root MSE = 5.9077

pov_inc Coef. Std. Err. t P>|t| [95% Conf. Interval]

mepi .9702335 .1230217 7.89 0.000 .7291022 1.211365


mpi 1.248595 .0042997 290.39 0.000 1.240168 1.257023
_cons 6.509978 .0895167 72.72 0.000 6.334519 6.685437

Diagnostic Tests
VIF - Test for Multicollinearity
. vif

Variable VIF 1/VIF

mepi 1.24 0.809540


mpi 1.24 0.809540

Mean VIF 1.24

Breusch-Pagan test for heteroskedasticity


. estat hettest

Breusch-Pagan / Cook-Weisberg test for heteroskedasticity


Ho: Constant variance
Variables: fitted values of pov_inc

chi2(1) = 1623.52
Prob > chi2 = 0.0000

78
OLS Regression with White’s Robust Standard Error
. regress pov_inc mepi mpi, robust

Linear regression Number of obs = 21960


F( 2, 21957) =55208.43
Prob > F = 0.0000
R-squared = 0.8294
Root MSE = 5.9077

Robust
pov_inc Coef. Std. Err. t P>|t| [95% Conf. Interval]

mepi .9702335 .1257849 7.71 0.000 .7236861 1.216781


mpi 1.248595 .0042987 290.46 0.000 1.24017 1.257021
_cons 6.509978 .0733188 88.79 0.000 6.366268 6.653688

Ramsey RESET
. estat ovtest, rhs

Ramsey RESET test using powers of the independent variables


Ho: model has no omitted variables
F(6, 21951) = 1123.60
Prob > F = 0.0000

2011
OLS Regression
. regress pov_inc mepi mpi

Source SS df MS Number of obs = 20591


F( 2, 20588) =67475.87
Model 3720424.11 2 1860212.05 Prob > F = 0.0000
Residual 567581.379 20588 27.5685535 R-squared = 0.8676
Adj R-squared = 0.8676
Total 4288005.49 20590 208.256702 Root MSE = 5.2506

pov_inc Coef. Std. Err. t P>|t| [95% Conf. Interval]

mepi 1.620471 .11617 13.95 0.000 1.392769 1.848173


mpi 1.54967 .0046922 330.26 0.000 1.540473 1.558867
_cons 5.021384 .0710631 70.66 0.000 4.882095 5.160674

79
Diagnostic Tests

VIF Test for Multicollinearity


. vif

Variable VIF 1/VIF

mepi 1.19 0.837223


mpi 1.19 0.837223

Mean VIF 1.19

Breusch-Pagan Test for Heteroscedasticity


. estat hettest

Breusch-Pagan / Cook-Weisberg test for heteroskedasticity


Ho: Constant variance
Variables: fitted values of pov_inc

chi2(1) = 6016.72
Prob > chi2 = 0.0000

OLS Regression with White’s Robust Standard Error


. regress pov_inc mepi mpi, robust

Linear regression Number of obs = 20591


F( 2, 20588) =27355.20
Prob > F = 0.0000
R-squared = 0.8676
Root MSE = 5.2506

Robust
pov_inc Coef. Std. Err. t P>|t| [95% Conf. Interval]

mepi 1.620471 .1181774 13.71 0.000 1.388834 1.852108


mpi 1.54967 .0074207 208.83 0.000 1.535125 1.564215
_cons 5.021384 .0832953 60.28 0.000 4.858119 5.18465

Ramsey RESET

. estat ovtest, rhs

Ramsey RESET test using powers of the independent variables


Ho: model has no omitted variables
F(6, 20582) = 3154.17
Prob > F = 0.0000

80
Multinomial Logistic Regression
. mlogit income mepi mpi pov_inc, baseoutcome (5)

Iteration 0: log likelihood = -18811.021


Iteration 1: log likelihood = -16230.526
Iteration 2: log likelihood = -16045.486
Iteration 3: log likelihood = -16034.281
Iteration 4: log likelihood = -16033.497
Iteration 5: log likelihood = -16033.479
Iteration 6: log likelihood = -16033.479

Multinomial logistic regression Number of obs = 20591


LR chi2(12) = 5555.08
Prob > chi2 = 0.0000
Log likelihood = -16033.479 Pseudo R2 = 0.1477

income Coef. Std. Err. z P>|z| [95% Conf. Interval]

Less_than_10_000
mepi 10.01881 1.617299 6.19 0.000 6.84896 13.18866
mpi -.2099127 .0422691 -4.97 0.000 -.2927587 -.1270667
pov_inc .1277527 .0252789 5.05 0.000 .0782071 .1772984
_cons 3.177927 .1677073 18.95 0.000 2.849227 3.506627

10_000_to_29_999
mepi 6.645693 1.617477 4.11 0.000 3.475495 9.81589
mpi -.1553998 .0422295 -3.68 0.000 -.2381681 -.0726314
pov_inc .0988872 .0252663 3.91 0.000 .0493661 .1484084
_cons 3.362128 .1674106 20.08 0.000 3.03401 3.690247

30_000_to_59_999
mepi 5.02759 1.622542 3.10 0.002 1.847465 8.207715
mpi -.0817469 .0428 -1.91 0.056 -.1656333 .0021395
pov_inc .059356 .025642 2.31 0.021 .0090986 .1096133
_cons 2.256063 .171851 13.13 0.000 1.919241 2.592885

60_000_to_99_999
mepi 3.031418 1.681195 1.80 0.071 -.2636648 6.3265
mpi -.0286496 .046204 -0.62 0.535 -.1192078 .0619085
pov_inc .0344875 .0279151 1.24 0.217 -.0202251 .0892
_cons .6446359 .1964966 3.28 0.001 .2595096 1.029762

100_000_and_over (base outcome)

B.3 Correlation Tests Results


B.3.1 Pearson Product Moment Correlation
2004
. pwcorr mepi mpi pov_inc, star (0.05)

mepi mpi pov_inc

mepi 1.0000
mpi 0.4364* 1.0000
pov_inc 0.4171* 0.9104* 1.0000

81
2011
. pwcorr mepi mpi pov_inc, star (0.05)

mepi mpi pov_inc

mepi 1.0000
mpi 0.4035* 1.0000
pov_inc 0.4079* 0.9308* 1.0000

B.3.2 Spearman Rank Test


2004 . spearman mepi mpi

Number of obs = 21960


Spearman's rho = 0.4328

Test of Ho: mepi and mpi are independent


Prob > |t| = 0.0000

. spearman mepi pov_inc

Number of obs = 21960


Spearman's rho = 0.4054

Test of Ho: mepi and pov_inc are independent


Prob > |t| = 0.0000
2011
. spearman mepi mpi

Number of obs = 20591


Spearman's rho = 0.4192

Test of Ho: mepi and mpi are independent


Prob > |t| = 0.0000

. spearman mepi pov_inc

Number of obs = 20591


Spearman's rho = 0.4046

Test of Ho: mepi and pov_inc are independent


Prob > |t| = 0.0000

82
B.3.3 Kendall Rank Test
2004
. ktau mepi mpi

Number of obs = 21960


Kendall's tau-a = 0.2854
Kendall's tau-b = 0.3286
Kendall's score = 68822394
SE of score = 1053548.830 (corrected for ties)

Test of Ho: mepi and mpi are independent


Prob > |z| = 0.0000 (continuity corrected)

. ktau mepi pov_inc

Number of obs = 21960


Kendall's tau-a = 0.2659
Kendall's tau-b = 0.3061
Kendall's score = 64107040
SE of score = 1053548.830 (corrected for ties)

Test of Ho: mepi and pov_inc are independent


Prob > |z| = 0.0000 (continuity corrected)
2011
. ktau mepi mpi

Number of obs = 20591


Kendall's tau-a = 0.2724
Kendall's tau-b = 0.3217
Kendall's score = 57737475
SE of score = 942620.997 (corrected for ties)

Test of Ho: mepi and mpi are independent


Prob > |z| = 0.0000 (continuity corrected)

. ktau mepi pov_inc

Number of obs = 20591


Kendall's tau-a = 0.2628
Kendall's tau-b = 0.3090
Kendall's score = 55704858
SE of score = 943316.090 (corrected for ties)

Test of Ho: mepi and pov_inc are independent


Prob > |z| = 0.0000 (continuity corrected)

83
C. Documentation

84
85
86
87
88
89

You might also like